Abstract. ñ Two diastereoisomers of the macrocyclic lactam alkaloid 3-hydroxycelacinnine with the (2R,3R) (1a) and (2R,3S) (1b) absolute configuration were synthesized by the alternative route involving macrocyclization with the epoxide ring opening by the terminal primary amine in the key step. Properly protected chiral (E)-epoxide precursors provided corresponding (2R,3R)-macrocycles in excellent yields (up to 85%) in the key macrocyclization step. However, due to the steric reasons, the macrocyclization of the corresponding (Z)-epoxides was unsuccessful. Inversion at C(3)-OH provided (2R,3S)-macrocycle. Synthesized (-)-(2R,3S)-3-hydroxycelacinnine (1b) was identical with the natural alkaloid.

Introduction. - Macrocyclic lactams derived from polyamines are of particular interest as synthetic targets for organic chemists due to the structural complexity and broad biological activity [1][2]. Séguineau et al. have isolated several novel hydroxylated spermidine alkaloids from the leaves of a New Caledonian Celastraceae, Pleurostylia opposita (Wall.) Merrill-Metcalf [3][4]. Their proposed structures for 3-hydroxycelacinnine (1) (originally named as 7-hydroxycelacinnine [3]), 7-hydroxypleurostyline (2), and 7-hydroxypleurocorine (3) are shown in Scheme 1. The presence of an OH group at the a-position to the lactam carbonyl group represents a new feature in such alkaloids. A biosynthetic pathway of their formation involving an (E)-epoxy precursor has been suggested [3][4] (Scheme 2). We are interested in structure verification and biosynthesis of these alkaloids.

Recently, we have described the synthesis of (±)-(2R*,3R*)-1a via stereoselective epoxide-ring opening with Mg(N₃)₂ and macrocyclization of the ditosylated diamine precursor with 1,4-dibromobutane promoted by Cs₂CO₃ in the two key steps [5] (Scheme 3).

A large difference between coupling constants (9.0 Hz vs. 1.2 Hz) as well as a significant difference between ¹H-NMR chemical shifts in the H-C(2)-C(3)-H moiety of the synthesized 1a and the natural 3-hydroxycelacinnine suggest that the proposed relative trans-configuration (2R*,3R*) for the natural alkaloid should be changed to the corresponding cis-configuration (2R*,3S*). Also, the same conclusion holds for 2 and 3, as well, since all three natural alkaloids had almost identical ¹H- and ¹³C-NMR data for this moiety [3][4]. Here we report the synthesis of both diastereoisomers of 3-hydroxycelacinnine (-)-1a and (-)-1b by the more efficient route according to the proposed biosynthetic pathway (Scheme 2) and confirm our hypothesis that (-)-1b is identical with the natural alkaloid.

Synthesis. ñ The optimized synthetic route to 1a is depicted in Scheme 4. After standard N-C coupling reactions starting from optically pure potassium (-)-(E)-3-phenylglycidate (5) [6] we have generated several N-protected precursors for the macrocyclization step. From several protective groups tested for the terminal primary amine (Boc, Ts, Troc, TFA), the best yields (up to 85%) were achieved with the TFA group, which is deprotected under the same conditions as the following macrocyclization step. Deprotection of other groups were accompanied either by complete (Ts, Boc) or by partial epoxide cleavage (Troc).

However, the corresponding TFA-protected (Z)-epoxide precursors 18 and 20 as well as separated intermediate 19 with a free primary amine group gave several side products and no desired macrocycle due to the steric reasons (Scheme 5). In the (Z)-epoxide (e.g. 19) the rotation of phenyl ring is restricted by the interaction with the neighboring amide, thus destabilizing the reactive conformation and blocking the nucleophilic approach of the amine. In contrast, the steric interactions with amide are absent in the (E)-epoxide and the molecule can be fold in the reactive conformation.

The desired (2R,3S)-macrocycle 15b was generated from the epimeric (2R,3R)-macrocycle 15a by mean of nucleophilic inversion at C(3) in the cyclic sulfamidate derivative 23 with the simultaneous N-protection and activation of C(3) towards nucleophilic displacement of oxygen. It was cleaved with the inversion by NaNO₂ followed by an acidic work-up to give the desired (2R,3S)-epimer 15b in 70% overall yield from 15a. Initial attempts to obtain cyclic sulfamidate via sulfamidite 22 (SOCl₂/Et₃N/CH₂Cl₂) has led to a small yield of the latter (15%). This was due to the facile aziridine ring formation via the participation of the neighboring unprotected nitrogen placed in anti-periplanar orientation in the intermediate 21 with the oxygen activated by the positively-charged residue (Scheme 6). However, generated in situ N,Në-thionyldiimidazole (Im/SOCl₂/CH₂Cl₂) has provided excellent yield (98%) of the cyclic sulfamidate 23 after oxidation of 22. Derivatization of C(3)-OH occurred faster than nitrogen N(1) due to the increased basicity of C(3)-OH next to the carbonyl group and steric hindrance of the nitrogen. Likewise, the attempted ditosylation of 15 has led exclusively to the aziridine ring formation. Furthermore, acylation in 16 can be carried out selectively at the C(3)-OH leaving hindered N(1)-atom intact (Scheme 4).

 
 

We have compared our synthesized (-)-(2R,3S)-3-hydroxycelacinnine with the 10-year old sample of the natural alkaloid which was kindly provided by Prof. Pascal Richomme (Université díAngers, France). They were identical by all means including 1D and 2D ¹H/¹³C-NMR, HPLC-MS/MS, UV, and optical rotation. In addition, Monte Carlo conformational search with MacroModel [7a,b] allowed us to calculate Boltzmann-averaged ¹H-NMR ³J coupling constants [7c] for the H-C(2)-C(3)-H moieties in 1a and 1b (9.7 and 1.2 Hz, respectively), which were in excellent agreement with the experimental values (9.6 and 1.2 Hz).

Conclusion. ñ We have confirmed the absolute configuration of the natural 3-hydroxycelacinnine by mean of total synthesis of its two diastereoisomers (-)-(2R,3R)- and (-)-(2R,3S)-3-hydroxycelacinnine 1a and 1b, respectively. Compound 1b was by all means identical with the natural alkaloid. In the light of our negative results with the macrocyclization of (Z)-epoxide precursors 18-20 of the (2R,3S)-macrocycle (Scheme 5), we conclude that the proposed biosynthetic pathway might work for the unnatural 1a (Scheme 2), but it is unlikely for the epimeric natural 1b. The alkaloid should be synthesized in vivo by an alternative pathway, for example, by the oxidation of the (-)-celacinnine in the presence of some mono-oxygenase enzymes.

[1] U. Bachrach, ¥Function of Naturally Occurring Polyamines¥, Academic Press, New York, 1973.

[2] A. Guggisberg, M. Hesse, in ¥The Alkaloids¥, Vol. 50, p. 219, Academic Press, New York, 1998.

[3] C. Séguineau, P. Richomme, J. Bruneton, J. Pusset, Helv. Chim. Acta 1992, 75, 2283.

[4] C. Séguineau, P. Richomme, J. Bruneton, P. Meadows, Heterocycles 1994, 38, 181.

[5] N. A. Khanjin, M. Hesse Helv. Chim. Acta 2001, 84, 1253.

[6] a) L. Thijs, J. J. M. Porskamp, A. A. W. M. van Loon, M. P. W. Derks, R. W. Feenstra, J. Legters, B. Zwanenburg, Tetrahedron 1990, 46, 2611; b) J. M. Lemmens, W. J. M. Blommerde, L. Thijs, B. Zwanenburg, J. Org. Chem. 1984, 49, 2231.

[7] a) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11, 440; b) http://www.schrodinger.com/macromodel2.html; c) C. A. G. Haasnoot, F. A. A. M. de Leeuw, C. Altona Tetraheron 1980, 36, 2783.