Proc. Nadl. Acad. Sci. USA Vol. 86, pp. 716-719, January 1989 Neurobiology Identification of the convulsant opiate thebaine in mammalian brain (radioimmunoassay/high-performance liquid chromatography/gas chromatography/mass spectrometry) HARUYOSHI KODAIRA*, CAROL A. LISEKt, IAN JARDINEt, AKIRA ARIMURAt, AND SYDNEY SPECTOR* *Roche Institute of Molecular Biology, Roche Research Center, Nutley, NJ 07110; tMayo Clinic, Rochester, MN 55905; and tTulane University, Hebert Center, Belle Chasse, LA 70037 Communicated by John J. Burns, October 12, 1988

ABSTRACT The convulsant opiate thebaine, an interme- 0.08 pmol. With this level of sensitivity we were able to detect diate of morphine biosynthesis, was purified from ovine brain a thebaine-like compound during the purification steps. to homogeneity by gel filtration and high-performance liquid Extraction and Purification Procedures. For purification of chromatography (HPLC) monitored by a radioimmunoassay. thebaine-like immunoreactive (ir) compound from ovine The immunoreactive material behaved identically to standard brain, Iyophilized ovine brain (2000 g) was pulverized, thebaine in two HPLC systems and was confirmed to be defatted with two times 3 vol of methylene chloride, and thebaine by combined gas chromatography/mass spectrome- dried. The defatted material was heated in 10 vol of 1 M acetic try. To our knowledge, the presence ofthebaine in mammalian acid containing 10 mM HCl at 95-100'C for 10 min, cooled on tissue has not been demonstrated previously. Codeine and ice, homogenized in a Polytron, and maintained at 40C morphine were also found to exist in ovine brain. The presence overnight. The material was centrifuged (10,000 x g, 30 min) of thebaine in ovine brain provides strong evidence that and the supernatant was subjected to ultrafiltration (Amicon morphine and codeine, in various mammalian tissues, are of YM-10) at 40C. The filtrate was Iyophilized. Step I (Fig. lA): endogenous origin and actually biosynthesized from a precur- Gel filtration at 40C on a 5.0 x 140 cm column of Sephadex sor. G-50 equilibrated with 1 M acetic acid. One-fourth of the extract was reconstituted with 150 ml of the eluent, centri- Recent evidence has focused attention on the existence of fuged, and applied to the column; flow rate was 15 ml/rnin endogenous opiate alkaloids in mammalian tissues (1-3). and 15-ml fractions were collected. The effluent was moni- These molecules may then constitute a component of a tored by UV absorbance at 280 nm (model 100-40 spectro- perhaps involving enkephalin, photometer, Hitachi). Of the total column volume, every 10 unique regulatory pathway fractions were combined and lyophilized. This gel filtration dynorphin, and endorphin, the more fully characterized procedure was repeated four times. To determine which opiate peptides. This notion is substantiated by the finding fractions from Sephadex G-50 column contain morphine- that the levels oftwo opiate alkaloids, morphine and codeine, related ir substances, fractions were then applied to a are not static but rather undergo rapid alterations following Sephadex G-15 column. The following three low molecular physiological stimulation (3, 4), whereas bioconversion of weight fractions were found to have the ir substance: frac- various opiate precursors can be demonstrated in vivo (1, 5) tions 121-130, 38 g; 131-140, 44 g; 141-150, 29 g. Step 2 (Fig. or in vitro (5, 6). A critical issue is to establish whether any 1B): Gel filtration on a 3.0 x 100 cm column ofSephadex G-15 of these intermediates actually exists in brain and further to equilibrated with 0.1 M pyridine/acetic acid, pH 4.7. One- determine whether they have biological activity in their own fifth offractions 121-130, from step 1, was reconstituted with right. We report here that one opiate, thebaine, a precursor 10 ml of buffer and applied to the column; flow rate was 1.07 in the biosynthesis of morphine and codeine that has con- ml/min and 11-ml fractions were collected. The effluent was vulsant properties, is present in ovine brain. monitored by UV detector (model 153, Altex) at 254 nm. Aliquots of each fraction were dried (Savant Speed-Vac concentrator), reconstituted in Dulbecco's phosphate-buf- METHODS fered saline (pH 7.4), and assayed with a RIA for the opiate Materials. Ovine brains (from Colorado Lamb, Denver, alkaloids. The ir fractions (bracket) from five successive runs CO) were used for the isolation of endogenous opiate alka- were pooled and dried. Step 3 (Fig. 2): One-eighteenth of the loids. (-)-Thebaine (Hoffmann-La Roche) crystallized from pooled material from step 2 was applied to a 0.4 x 25 cm methanol was used for radioimmunoassay (RIA), high- Merck LiChrosorb RP18 column (5-.um particles), using a performance liquid chromatography (HPLC), and gas chro- Beckman model 322 solvent delivery system, eluted at a rate matography/mass spectrometry (GC/MS) analysis as a stan- of 1.45 ml/min with 0.1 M pyridine/acetic acid, pH 4.7, dard. (-)-Codeine was from Merck. Morphine hydrochloride containing 12% acetonitrile (vol/vol), and then flushed with and '25I-labeled morphine (90-150 Ci/mmol; 1 Ci = 37 GBq) 0.2 M pyridine/acetic acid, pH 4.7, containing 75% acetoni- were from Hoffmann-La Roche. Other reagents were pur- trile (vol/vol). The elution was monitored by electrochemical chased from Baker or Sigma. (EC) detector (model LC-4A, Biochemical Analytical Sys- RIA. A rabbit antiserum that was raised against 3- tems) at an oxidation potential of 500 mV vs. a Ag/AgCl carboxymethylmorphine conjugated to bovine serum albu- electrode (6). Aliquots offractions were assayed by RIA. The min (7) and 1251-labeled morphine as tracer were used in the two major ir fractions (brackets) from 18 successive runs RIA. The sensitivity of the assay to thebaine [IC50 = 0.58 + were pooled and dried. Step 4 (Fig. 3): The combined second 0.04 pmol (mean ± SEM)] is about 1/10th that of morphine peak (bracket) from the 18 runs in step 3 was applied to a 0.46 or codeine, both of which react equally well with the x 25 cm Whatman Partisil 10 SCX ion-exchange column and antiserum (6). The detectiori limit of the assay for thebaine is eluted isocratically at a rate of 1 ml/min with 0.2 M acetic acid/pyridine, pH 3.5. The effluent was monitored by EC detector at 800 mV. Aliquots of fractions were assayed by The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviations: ir, immunoreactive; EC, electrochemical.

716 Downloaded by guest on September 30, 2021 Neurobiology: Kodaira et al. Proc. Natl. Acad. Sci. USA 86 (1989) 717

500- -I, 2.0- A LU C LE ii I Z > Z c 00 250-- co m0 1.0ki co ii LU F

M E OL IU CL l01r ;, _ 0 50 100 150 200 0.32 B A 0 50 70 90 110 130 LU f % C! RETENTION TIME (min) Zr0EC-

cr cj 0. 16- FIG. 3. Ion-exchange HPLC chromatogram of thebaine-like ir 0CO) compound from reverse-phase HPLC. Effluent was monitored by m EC detection and RIA. Shaded bar graphs represent thebaine immunoreactivity. The arrow indicates the position of standard thebaine in the HPLC system. z 4F It= o (vol/vol) or dimethyl sulfoxide to remove any standard

aO C c retained in the system. After washing, immunoreactivity was oL c0 25 50 75 100 not detected at the retention times of the standards. FRACTION NUMBER GC/MS Procedure. The HPLC solvent blank at the reten- tion time of thebaine, the putative thebaine from ovine brain FIG. 1. Gel filtration chromatograms and morphine-related (30 pmol), and standard thebaine (35 pmol), which were immunoreactivity: Sephadex G-50 of ovine brain extract (A) and collected at step S in Fig. 4, were used for this analysis. Sephadex G-15 of fractions 121-130 from Sephadex G-50 (B). Between each analysis, a solvent (acetonitrile) blank was also Effluent was monitored by UV detection and RIA for morphine. used to eliminate any possible contamination of the endog- Shaded bar graphs represent morphine-related immunoreactivity. enous sample with standard sample. No trace of peaks RIA. The position of the ir peak was identical to that of corresponding in m/z values ofmajor ions in thebaine spectra standard thebaine (arrow). The fractions between 87 and 93 were observed in any of the blanks analyzed. The GC/MS min were combined and evaporated to dryness. Step 5 (Fig. system used was a VG Masslab 30-253 quadrupole mass 4): The ir peak from step 4 was applied to the same RP18 spectrometer (electron-ionization mode at 25 eV and 150 ,uA) column as in step 3 and eluted at 1.45 ml/min with 0.1 M interfaced to a Hewlett-Packard 5890 gas chromatograph. pyridine/acetic acid, pH 4.7, containing 12% acetonitrile, GC column: Hewlett-Packard crosslinked methyl silicone- giving a detectable EC response peak at 750 mV and an ir fused silica capillary, 25 m x 0.31 mm i.d.; film thickness, peak (total, 45 pmol). The ir peak, coeluted with standard 0.17 Am. Injection port and transfer line: 200'C and 220'C. thebaine (arrow), was evaporated to dryness and a portion GC oven conditions: 50'C held for 2 min; 10'C/min to 2250C; (30 pmol) was used for GC/MS analysis. Portions of the 2250C held for 2 min. Carrier gas: He. HPLC solvent blank and standard thebaine that were passed through the reverse-phase HPLC system (Fig. 4) were also used for GC/MS analysis (see below). During the HPLC RESULTS purification steps, standard thebaine, codeine, or morphine Thebaine-like ir compound was purified from an acetic was assayed following each fractionation step. Following acid-containing HCl extract of ovine brain by gel filtration, application of the standard, the injection system and the reverse-phase (RP18), and ion-exchange HPLC systems. HPLC columns were washed with methanol and/or chloro- Fractions were assayed with a RIA to detect morphine- form plus multiple injections of 30% acetic acid/methanol 500- °E50

(I) z > 3 -- -0 = 0 250- |,,_

LULO= CO 0 13lar 0L Z - w 0L- z z 3- z 30 2- - m 20- 1 wjE E1 I a 10 - o 02 .Z 0 10 20 30 40 50 0* 0 10 20 30 40 5 0 RETENTION TIME (min) RETENTION TIME (min) FIG. 2. Reverse-phase HPLC chromatogram of morphine- FIG. 4. Reverse-phase HPLC chromatogram of thebaine-like ir related ir compounds from Sephadex G-15. Effluent was monitored compound from ion-exchange HPLC. Effluent was monitored by EC by EC detection and RIA. Shaded bar graphs represent immuno- detection and RIA. Shaded bar graphs represent thebaine immuno- reactivity. The second major ir peaks (bracket) were calculated using reactivity. Trace A, HPLC solvent blank; trace B, EC response peak thebaine as a standard. The arrow indicates the position of standard of standard thebaine when 65 pmol of the standard was applied to the thebaine in the HPLC system. The first major (bracket) and the other HPLC system (arrow). Portions of these samples were used for minor ir peaks were calculated using morphine as a standard. GC/MS analysis (see Fig. 5). Downloaded by guest on September 30, 2021 718 Neurobiology: Kodaira et al. Proc. Natl. Acad Sci. USA 86 (1989) related ir substances, including morphine, codeine, and 5). From these results, we conclude that the isolated com- thebaine. pound is thebaine. The first ir peak at step 3 (Fig. 2) was also Figs. 1-4 show the purification procedures ofthebaine-like purified by RP18 and ion-exchange columns using different ir compound from ovine brain. The ovine brain acid extract solvent systems (see ref. 6), resulting in the isolation of was first subjected to Sephadex G-50 filtration (Fig. lA). The codeine (total, 64 pmol) and a trace of morphine. RIA was not performed at this stage because of the high concentration of acid in each fraction, which interfered with DISCUSSION the assay. The low molecular weight material from the Sephadex G-50 filtration was applied to a Sephadex G-15 In this paper we report the presence of the alkaloid thebaine column, and a broad region of morphine-related ir com- in mammalian tissue. In addition to serving as an intermedi- pounds was found (Fig 1B). Application of this ir fraction to ate in the synthesis of codeine and morphine (1, 6), thebaine a RP18 column resulted in two main ir peaks (Fig. 2). The first may also exert some distinct physiological effect of its own. peak was similar to the position where standard morphine Higher concentrations of codeine and morphine can be and codeine eluted. Morphine and codeine cannot be sepa- detected in tissues after boiling and acid hydrolysis with 10% rated using this solvent system. The second peak eluted in the HCl, since these alkaloids exist in a conjugated form or bound same position as standard thebaine. The second peak was to protein (3). As thebaine is unstable under acid hydrolysis, then applied to an ion-exchange column for further purifica- this procedure was omitted. The codeine and morphine tion. It was found to have an identical retention time as that isolated in this study were probably free in tissue. Thebaine of standard thebaine (Fig. 3). The thebaine-like ir peak was can bind to plasma protein and albumin (9); therefore, it is rechromatographed on the same RP18 column used in step 3 possible that thebaine might exist in tissue bound to proteins. (Fig. 2). A single EC response peak associated with immuno- In the poppy plant, Papaver somniferum, the morphine- reactivity occurred where standard thebaine eluted (Fig. 4). type alkaloid thebaine is biosynthesized from two molecules The total amount isolated was 45 pmol. Portions of this of L-tyrosine (10) by way of several intermediates, such as endogenous putative thebaine, and also standard thebaine norcoclaurine, coclaurine (11, 12), reticuline (13), and salu- and HPLC solvent blank collected at step 5 in Fig. 4, were taridine (14). It has recently been shown that norcoclaurine then subjected to GC/MS analysis (Fig. 5). The electron- and coclaurine are true precursors of reticuline and thebaine ionization mass spectrum of standard thebaine has a base (11, 12), though tetrahydropapaveroline (THP, norlaudano- peak corresponding to the molecular ion at m/z 311 and two soline) had long been assumed to be the precursor (15, 16). major fragment ions at m/z 296 and m/z 255. These fragment The possible role of THP as a precursor in the biosynthesis ions result from loss of a methyl group and the N-methyl ofmorphine in mammals is unclear. THP, which is formed by bridge with hydrogen transfer (C3H6N) (8), respectively. the condensation between dopamine and an aldehyde derived Standard thebaine and the endogenous putative thebaine from a second dopamine molecule (17), has been identified in eluted at the same GC oven temperature (192°C) and reten- the brain of rats receiving L-3,4,-dihydroxyphenylalanine tion time (16.2 min), and all three ofthe above diagnostic ions (L-dopa) (18) as well as in the urine of parkinsonian patients plus numerous other characteristic lower mass fragment ions who have been given L-dopa (19). However, it has not been occurred in both samples with similar relative intensities (Fig. detected in normal mammalian tissue nor has it been shown to be a precursor for reticuline in mammalian tissue. In plants 186le90 A 31JImM+ (20) morphine is formed from thebaine by way of two >- 8o pathways: (i) from codeine or (it) from oripavine. In mam- malian tissue it has been shown that salutaridine can be z (M-Cv+3) formed from reticuline (5) and that the concentration of 30- 2e6 morphine and codeine increased following administration of thebaine or salutaridine (1). The possibility that the thebaine z SO- found in the ovine brain is of a dietary origin cannot be excluded since the brain was obtained from a slaughterhouse w (M-QCJ)+ and we were unable to test for the presence ofthebaine in the 16 I22 O2m food. However, we have recently shown that microsomes ao from various mammalian tissues in the presence of cofactors 126 140 160 180 2k0 226 240 290 286 386 312 34 can convert thebaine to morphine; the biosynthetic pathway m/z was similar to that that exists in plants (6). Therefore, the presence of thebaine as well as codeine and morphine in ovine brain provides strong evidence that 31 1 Ma 166 B morphine and codeine, which had been identified in mam- I-O90 >. (M-CH mals, are of endogenous origin and biosynthesized from the is precursor in mammalian tissue. so-Z 70- Thebaine exerts profound physiological and pharmacolog- Z860 I\Z9JCH3 ical effects. It has slight analgesic and depressant effects and does not elicit physical dependence (21, 22). However, Cle (M-C~s1) thebaine can stimulate the central nervous system and cause -J 30 wC 2 e convulsions (21-25). Morphine and codeine also produce 20 convulsion but only at high doses (21, 23, 26), whereas at low 5 a doses, morphine can exert an anticonvulsant action A. I k l --- (26). 12 14 16 2l 8 226 246 266 28 300 320 Thebaine can reverse some of the effects of morphine (27, m/z 28). Electrophysiological studies indicate that the spasmodic effect of thebaine was different from that of morphine and FIG. 5. Electron-ionization mass spectra recorded for the GC/MS analysis: putative thebaine isolated from ovine brain (A) and codeine but was similar to that of strychnine (29). Thebaine standard thebaine (B). The HPLC solvent blank at the retention time has been shown to inhibit binding to the glycine receptor and of thebaine, the putative thebaine from ovine brain (30 pmol), and the -y-aminobutyric acid receptor (30). Both receptors seem to standard thebaine (35 pmol), which were collected as shown in Fig. be involved in the convulsive properties of the opiates (31, 4, were used for this analysis. 32). The finding of endogenous thebaine in mammalian brain Downloaded by guest on September 30, 2021 Neurobiology: Kodaira et al. Proc. Natl. Acad. Sci. USA 86 (1989) 719

tissue opens further areas of investigation as to the role of 14. Barton, D. H. R., Kirby, G. W., Steglich, W., Thomas, thebaine and its relationship to codeine and morphine. G. M., Battersby, A. R., Dobson, T. A. & Ramuz, H. (1965) J. Chem. Soc., 2423-2438. 15. Battersby, A. R., Foulkes, D. M. & Binks, R. (1965) J. Chem. This work was supported in part by National Institutes of Health Soc., 3323-3332. Grant GM 32928. 16. Haslam, E. (1986) Nat. Prod. Reports 3, 217-248. 17. Holtz, P., Stock, K. & Westermann, E. (1964) Nature 1. Donnerer, J., Oka, K., Brossi, A., Rice, K. C. & Spector, S. (London) 203, 656-658. 18. Turner, A. J., Baker, K. M., Algeri, S., Frigerio, A. & (1986) Proc. Natl. Acad. Sci. USA 83, 4566-4567. Garrattini, S. (1974) Life Sci. 14, 2247-2257. 2. Weitz, C. J., Lowney, L. I., Faull, K. F., Feistner, G. & 19. Sandler, M., Carter, S. B., Hunter, K. R. & Stern, G. M. Goldstein, A. (1986) Proc. Natl. Acad. Sci. USA 83, 9784- (1973) Nature (London) 241, 439-443. 9788. 20. Brochmann-Hanssen, E. (1985) in The Chemistry and Biology 3. Donnerer, J., Cardinale, G., Coffey, J., Lisek, C. A., Jardine, ofIsoquinoline Alkaloids, eds. Philipson, J. D., Roberts, M. F. I. & Spector, S. (1987) J. Pharmacol. Exp. Ther. 242, 583-587. & Zenk, M. H. (Springer, Berlin), pp. 229-239. 4. Haberman, F., Lavicky, J., Marcum, E. & Spector, S. (1988) 21. Krueger, H., Eddy, N. B. & Sumwalt, M. (1943) The Phar- FASEB J. 2, 5606 (abstr.). macology of the Opium Alkaloids, Parts I and 2 (U.S. 5. Weitz, C. J., Faull, K. F. & Goldstein, A. (1987) Nature Government Printing Office, Washington, DC). (London) 330, 674-677. 22. A WHO Advisory Group (1980) Bull. Narcot. 32, 45-54. 23. Corrado, A. P. & Longo, V. G. (1961) Arch. Int. Pharma- 6. Kodaira, H. & Spector, S. (1988) Proc. Nall. Acad. Sci. USA codyn. Ther. 132, 255-269. 85, 1267-1271. 24. Gilbert, P. E. & Martin, W. R. (1975) J. Pharmacol. Exp. Ther. 7. Gintzler, A. R., Mohacsi, E. & Spector, S. (1976) Eur. J. 192, 538-541. Pharmacol. 38, 149-156. 25. Tortella, F. C., Cowan, A. & Adler, M. W. (1984) Neuro- 8. Wheeler, D. M. S., Kinstle, T. H. & Rinehart, K. L., Jr. pharmacology 23, 749-754. (1967) J. Am. Chem. Soc. 89, 4494-4501. 26. Frenk, H. (1983) Brain Res. Rev. 6, 197-210. 9. Misra, A. L., Pontani, R. B. & Mule, S. J. (1974) Xenobiotica 27. McClane, T. K. & Martin, W. R. (1967) Int. J. Neurophar- 4, 17-32. macol. 6, 325-327. 10. Spenser, I. D. (1968) in Comprehensive Biochemistry, eds. 28. Messiha, F. S. & Krantz, J. C., Jr. (1973) Arch. Int. Phar- Florkin, M. & Stotz, E. H. (Elsevier, Amsterdam), Vol. 20, macodyn. Ther. 206, 90-93. pp. 231-413. 29. Longo, V. G. (1962) in Rabbit Brain Research (Elsevier, 11. Stadler, R., Kutchan, T. M., Loeffler, S., Nagakura, N., Amsterdam), Vol. 2. Gassels, B. & Zenk, M. H. (1987) Tetrahedron Lett. 28, 1251- 30. Goldfinger, A., Muller, W. E. & Wollert, U. (1981) Gen. 1254. Pharmacol 12, 477-479. 12. Loeffler, S., Stadler, R., Nagakura, N. & Zenk, M. H. (1987) 31. Dingledine, R., Iversen, L. L. & Breuker, E. (1978) Eur. J. J. Chem. Soc. Chem. Commun. 15, 1160-1162. Pharmacol 47, 17-27. 13. Battersby, A. R., Binks, R., Francis, R. J. & Ramuz, H. 32. Werz, M. A. & Macdonald, R. L. (1982) Brain Res. 236, 107- (1964) J. Chem. Soc., 3600-3610. 119. Downloaded by guest on September 30, 2021