Provided by total synthesis, endoperoxides 18, 20, and 22 underwent intramolecular oxymercuration-demercuration leading respectively to formation of an isomeric tetracycle, (1aS, 3S, 5aS, 6R, 8aS, 9R, 12S)-10-deoxo-13-carbaartemisinin (19), (+)-10-deoxo-13-carbaartemisinin (21), and (+)-13-carbaartemisinin (4). Structure assignment to 19 and 21 was based on single-crystal X-ray crystallographic analysis. Tricyclic endoperoxide 20 was converted to methyl and benzyl ethers 23 and 24 and reduced to saturated analog 25 which was also converted to ethers 26 and 27. In vitro antimalarial screening of both tri- and tetracyclic analogs was conducted using the W-2 and D-6 clones of Plasmodium falciparum. Neither target 4 nor 21 displayed substantial antimalarial potency in vitro against P. falciparum, but the diastereomeric peroxide 19 possessed good antimalarial potency in vitro. Tricyclic analogs were uniformly impotent. Iron(II) bromide-promoted rearrangement of 21 gave, in 79% yield, the unique tetracyclic alcohol 35, while 19 provided ring-opened cyclohexanone 41 (39%) along with the tricyclic epoxide 42 (20%). Neither 41 nor 42 possessed in vitro antimalarial activity, suggesting that epoxide-like intermediates are not responsible for the mode of action of this subclass of antimalarials. Rearrangement of 10-deoxoartemisinin (43) with FeBr2 gave a major product (79%) not encountered in the rearrangement of artemisinin that resulted from unraveling of the tetracyclic system cyclohexanone 46. Minor amounts of 1,10-dideoxoartemisinin (49) (8%) were also produced in this reaction.

Novel 3- and 9-substituted analogs (4-19) of 10-deoxoartemisinin, 3, were prepared from the corresponding known lactones by one-pot reduction with sodium borohydride and boron trifluoride etherate. Reproducibility problems associated with this heterogeneous reaction were encountered on small reaction scales, and thus alternative methodology was sought for this reduction. Conversion of the lactones to tetrahydropyrans via the corresponding intermediate lactols was made more reproducible using a two-step sequence involving low-temperature reduction with diisobutylaluminum hydride followed by deoxygenation with boron trifluoride etherate in the presence of triethylsilane. In this manner, 10-deoxoartemisinin (3) could be obtained from artemisinin (1) in greater than 95% overall yield. All analogs were tested in vitro against W-2 and D-6 strains of Plasmodium falciparum. Several of the analogs were much more active than the natural product (+)-artemisinin (1) or 10-deoxoartemisinin (3). Conventional structure-activity relationships are discussed in relation to the bioassay data.

A novel class of artemisinin analogs, N-alkyl-11-aza-9-desmethylartemisinins 17-29, were synthesized via ozonolysis and acid-catalyzed cyclization of precursor amides 5-16. These amides were prepared through condensation of an activated ester of the known intermediate acid 2 with the corresponding primary amine. The analogs were tested in vitro against W-2 and D-6 strains of Plasmodium falciparum and found in some cases to be more active than artemisinin. A comparison of the in vitro testing methods of Milhous and Makler was conducted and gave similar relative antimalarial activities for these artemisinin analogs. Log P values were determined for most of the compounds, but no apparent correlation between log P and in vitro activity was found.