The time course of the levels of artemisinin, its biosynthetic precursors and the biosynthetically related sesquiterpenes was monitored during a vegetation period of Artemisia annua plants of different geographical origin. Considerable differences in contents of artemisinin and its direct precursors artemisinic acid and dihydroartemisinic acid were found between these A. annua's. For the first time the A. annua plants of different geographical origin were found to belong to different chemotypes. A chemotype with a high artemisinin level was found to have also a high dihydroartemisinic acid level but a relatively low artemisinic acid level. Reversibly, a chemotype with low levels of artemisinin and dihydroartemisinic acid contained a high artemisinic acid level. Artemisinic acid is considered to be the direct precursor of dihydroartemisinic acid in the biosynthetic pathway of artemisinin. The observed accumulation of artemisinic acid in one of the A. annua chemotypes may indicate the presence of a rate-limiting step in the biosynthetic pathway of artemisinin. The enzymatic reduction of artemisinic acid into dihydroartemisinic acid is probably a "bottle neck" in the biosynthetic pathway of artemisinin in varieties with high artemisinic acid and consequentially low artemisinin levels. After a night-frost period, the level of artemisinin was increased, in the Vietnamese A. annua plants, while the dihydroartemisinic acid level was decreased. This phenomenon is in accordance with our hypothesis that stress triggers the conversion of dihydroartemisinic acid to artemisinin. It is suggested that the presence of high levels of dihydroartemisinic acid may be an adaptation to stress conditions (e.g., night-frost), during which relatively high levels of 1O2 are formed. Dihydroartemisinic acid gives the plant protection by reacting with these reactive oxygen species yielding artemisinin as stable end-product.

We determined the cytotoxicity of some artemisinin derivatives against EN2 tumor cells using the MTT assay. Artemisinin (1) was clearly more cytotoxic than deoxyartemisinin (2), which lacks the endoperoxide bridge. Ether-linked dimers of dihydroartemisinin with defined stereochemistry were found to differ in the extent of cytotoxic effect on EN2 cells. The nonsymmetrical dimer (3) was more cytotoxic than the symmetrical dimer (4). The nonsymmetrical dimer of dihydrodeoxyartemisinin (5) lacking the endoperoxide bridges was also effective in the MTT assay, although less cytotoxic than 3 and 4. Similarly, the symmetrical dimer (6) was less effective than 5. Epoxides of artemisitene also showed that stereochemistry was an important factor for cytotoxicity. The results suggested that the endoperoxide bridge was not crucial for cytotoxicity to the tumor cells, but contributed to the cytotoxic effect apparently exerted by the ether linkage of the dimers. Flow cytometry data indicated that the dimers 3 and 4 caused an accumulation of the cells in the G1-phase of the cell cycle. In contrast, artemisinin (1) caused a slight increase of S-phase cells.

This study deals with the cytotoxicity of helenanolide-type (10 alpha-methylpseudoguaianolide) sesquiterpene lactones. We determined the influence of substitution patterns on the toxicity of 21 helenanolides to a cloned Ehrlich ascites tumor cell line, EN2. Within a series of helenalin esters, the acetate (2) and isobutyrate (3) were more toxic than helenalin itself (1). Esters with larger acyl groups (tiglate 4 and isovalerate 5) exhibited a decreased toxicity compared with the parent alcohol (1). Similar relationships were observed between the 6,8-diastereomer of helenalin, mexicanin I (6) and its acetate (7) and isovalerate (8). In contrast, cytotoxicity within a series of 11 alpha, 13-dihydrohelenalin esters (9-12) was shown to be directly related to the size and lipophilicity of the ester side chain, dihydrohelenalin (9) being the least toxic compound in this group. Investigation of several 2,3-dihydrohelenalin derivatives (13-21) with 2 alpha-hydroxy-4-oxo- and 2 alpha,4 alpha-dihydroxy- or -O-acyl-substituted cyclopentane rings (arnifolins and chamissonolides, respectively), for which no pharmacological data have been reported so far, revealed further interesting influences of the substitution pattern on cytotoxicity. The results may be interpreted in terms of lipophilicity and steric effects on the accessibility of the reactive sites considered responsible for biological activity.

We determined the in vitro cytotoxic activity of the sesquiterpene lactone endoperoxide artemisinin (1) and some chemically prepared derivatives, which have been found to display cytotoxicity to cloned murine Ehrlich ascites tumour (EAT) cells and human HeLa cells and against murine bone marrow using a clonogenic assay for committed progenitor cells of the granulocyte-monocyte lineage (CFU-GM assay). Comparing artemisinin (1) to deoxyartemisinin (2), the endoperoxide group appeared to play a role in cytotoxicity to CFU-GM cells. Dimers of dihydroartemisinin and dihydrodeoxyartemisinin revealed that the stereochemistry of the ether linkage of the dimers was a more important determinant for this cytotoxic activity. The nonsymmetrical dimer of dihydroartemisinin (3) and the corresponding endoperoxide-lacking dimer of dihydrodeoxyartemisinin (5) were equally cytotoxic to CFU-GM cells. Despite the differences between both systems, it may be stated that most compounds displayed higher cytotoxicity to CFU-GM cells than to EAT cells. Dimers of dihydroartemisinin (3, 4) were selected as potential antitumour compounds and subjected to the National Cancer Institute drug-screening programme consisting of about sixty human cancer cell lines derived from nine different tissues. Both compounds displayed the same specific cytotoxicity pattern. Throughout the screen dimer 3 was more active than 4.