The chewing of the leaves of Catha edulis Forsk (Khat) by the natives of several Asian and African countries to provide rapid stimulation is extremely prevalent¹ and has been considered to be a serious problem of drug dependence not unlike that associated with amphetamine². In fact, on the basis of the observations of Eddy et al.³ the United Nations Narcotics Laboratory undertook research on the chemistry of Khat and its components².

Earlier in this century (+)-norpseudoephedrine, a CNS-active compound, was identified among the basic alkaloid components of Khat⁴. Later investigations^5-8 revealed the presence of other bases, but the isolation and characterization of the main constituent has only been reported within the last few years by the United Nations Narcotics Laboratory^9,10. Specifically, the presence of (-)-α-aminopropiophenone⁹ and its absolute configuration¹⁰, as well as the presence of 3,6-dimethyl-2,5-diphenylpyrazine⁹ were reported. Since this information is only available in the United Nations Documents which are somewhat hard to come by¹¹, it will be briefly summarized here. Extraction of freeze-dried plant material with methanol, followed by separation of the nonpolar and weakly basic species and purification by acid-base extractions gave an odiferous yellow oil which, after treatment with oxalic acid, gave a solid: mp 157-160°C⁹. This compound was identified as α-amino- propiophenone oxalate by its UV, IR, ¹H-NMR, and mass spectra⁹. In another reports the absolute configuration of the plant extract was deduced from comparison of the CD curve observed for a separately isolated sample of oxalate salt with those of model compounds. It was concluded¹⁰ that natural α-aminopropiophenone has the S configuration, which is biogenetically consistent with the configurations of both (+)-norpseudo- ephedrine and (-)-norephedrine at the corresponding asymmetric center.

These reports do, however, contain some puzzling results. The stereospecific synthesis of (-)-α-aminopropiophenone by bromination of propiophenone followed by the Gabriel synthesis is claimed¹⁰. Indeed, the melting point (175°C) reported for the hydrochloride salt of the synthetic product differs substantially from that reported previously for racemic α-amino- propiophenone hydrochloride (184°C (12), and 187°C¹³ and is in agreement with the melting point reported by Takamatsu (175-176°C¹⁴) for (+)-α- aminopropiophenone hydrochloride. Furthermore, it was reportedly that when this salt was converted to an amine oxalate, a solid with a melting point of 172-175°C, which showed no melting point depression with natural cathinone oxalate, was obtained.

These results seem to suggest that the stereospecific synthesis of (-)-α-aminopropiophenone hydrochloride from propiophenone had been accomplished without the benefit of optically active reagents, solvents, or resolution. Other apparent inconsistencies appear in the literature relating to α-aminopropiophenone. For example, the report¹⁴ that treatment of α-aminopropiophenone with ethanolic hydrochloric acid gave (+)-α-amino-propiophenone hydrochloride, mp 175-176°C, would appear to be an error presumably the l-mandelate salt of (+)-α-aminopropiophenone was treated with ethanolic hydrochloric acid.

In the light of increased interest in expanding the pharmacological and medical investigations of the major constituents of Khat² and because known methods of resolution of racemic α-aminopropiophenone were inadequate for the preparation of gram quantities of the required antipodes of α-aminopropiophenone, we undertook the development of an improved synthetic pathway to optically active α-aminopropiophenone. At the same time, having pure α-aminopropiophenone and its optical antipodes in hand, we were able to establish some of the properties and chemical behavior of α-aminopropiophenone and of its salts and to explain some of the literature discrepancies.

Results and Discussion

Attempts to resolve (-)-α-aminopropiophenone via the mandelate salt by following the reported procedures^7,14 yielded only minute quantities of product in spite of the care taken to exclude light. Somewhat better results were obtained with tartaric acid. Thus, (+)-α-aminopropiophenone crystallized with (-)-tartaric acid and the (-) enantiomer crystallized with (+)-tartaric acid. For both antipodes the yields were small, probably due to the strong tendency of α-aminopropiophenone (1) to cyclize to 3,6-dimethyl-2,5-diphenyldihydropyrazine (3), with subsequent oxidation to 3,6-dimethyl-2,5-diphenylpyrazine (2) when it is not stabilized by the presence of strong acids¹³.

Consequently, an alternative route to optically active cathinone was developed. Norephedrine was resolved in high yield into its (+) and (-) antipodes with O,O-dibenzoyl- d-tartaric acid¹⁵. Each enantiomer was converted to its N-formyl derivative and oxidized with chromium trioxide in pyridine. Hydrolysis with 20% hydrochloric acid at 40°C gave optically pure α- aminopropiophenone hydrochloride without racemization. In this manner there was obtained, from racemic norephedrine, (-)-α-aminopropiophenone in 39% overall yield and the (+) enantiomer in 40% overall yield. It should be noted that use of an N-acetyl blocking group in the oxidation of norephedrine to α-aminopropiophenone was unsuitable because removal of the acetyl group led to racemization (~60%). Similarly, if deprotection of N-formyl-α-aminopropiophenone was carried out with weaker acid (<20% HCl) and at higher temperature (>40°C), racemization was observed.

An approach to the stereospecific synthesis of (S)-cathinone from (S)-N-(methoxycarbonyl)alanine has been recently published¹⁷. The Friedel-Crafts yield was reported to be 50-60%; assuming a 90% yield in removal of the N-methoxycarbonyl protecting group, the overall yield obtained would be slightly higher than that obtained by our method. However, whereas our approach leads to both enantiomers on using a natural (and therefore relatively inexpensive) tartaric acid derivative, preparation of (R)- cathinone by the procedure of McClure et al¹⁷ requires nonnatural alanine.

Surprisingly, the melting points obtained by us for racemic and optically active α-aminopropiophenone hydrochloride were practically the same (racemic, mp 190-191°C; optically active, mp 189-191°C). Furthermore, our melting points for the optical antipodes were substantially higher than those previously reported (175-176°C)¹³. Our experiments suggest that the low melting point was observed when the samples were insufficiently dried.

Because it had been found that optically active α-aminopropiophenone racemized readily in the absence of strong acids¹⁸, it occurred to us that the plant extract obtained by using dilute acetic acid at room temperature might have racemized to some extent during the extraction procedure and/or while present as the salt of a weak acid such as oxalic acid. The hydrochloride salts of (-)-α-aminopropiophenone and of the racemate were therefore converted to the oxalate salts. The optically active and racemic oxalate salts had very similar melting points (optically active, mp 173-175°C; racemic, mp 172-173°C), as we had found for the hydrochloride salts. In addition, they did not exhibit melting point depression when admixed. Thus, although optical resolution by direct crystallization of mixtures of interconverting enantiomers is possible in principle¹⁹, the strong tendency of α-aminopropiophenone to racemize and/or cyclize when it is not stabilized by the presence of a strong acid^13,18 may account for the apparent identity of the oxalate salt of the plant extract and of synthetic α-aminopropiophenone observed¹⁰. Clearly, in this case the melting points cannot serve as criteria for optical purity.

Conclusions

1. The optical antipodes of α-aminopropiophenone can be prepared in high yield by oxidation of optically active N-formylnorephedrine, followed by deblocking of the amino group. The resolution of nor-ephedrine is best carried out by using O,O-dibenzoyltartaric acid.
2. Melting points cannot be used as criteria of the optical purity of cathinone derivatives because of the (a) similarity of the melting points of the racemic and optically active compound and (b) lack of melting point depression upon admixing of opposite antipodes and/or racemic and optically active compounds.
3. The cathinone free base racemizes and dimerizes readily in a hydroxylic medium. Similar behavior is observed for solutions of the oxalate salt but at a somewhat reduced rate; therefore, the plant extract of Schorno¹⁰ may have been partially racemized.
4. The cathinone free base is fairly stable as a dilute solution in nonpolar, nonhydroxylic media, and, consequently, strong acid salts of optically active cathinone can be interconverted without racemization.