In recent years, the clinical management of malaria has become more challenging due to progressive resistance to drugs such as chloroquine, which was once regarded as the mainstay of malaria chemotherapy.1
The emerging resistance has fueled a search for new effective antimalarial agents. Exploration of natural products is a particularly attractive avenue of pursuit, in part because nature has provided the leads for the most successful drugs for the treatment of malaria during the past 60 years (quinine, quinidine, and/or their analogs), including, most recently, artemisinin. The identification of artemisinin as the main bioactive ingredient of Artemisia annua
and the subsequent elaboration of the unique mechanism of action underlying its antiplasmodial activity1b
have focused new attention on endoperoxides.2
Endoperoxides are abundant in nature2a
and display a broad range of pharmacological properties including antifungal, cytotoxic, antiviral, antitrypanosomal, and antiplasmodial activities.1a
Herein, we report the isolation, structure elucidation, and antiplasmodial activity of (+)-(3R
)-one (okundoperoxide, 1
, ), a new and skeletally unique bicyclofarnesyl sesquiterpene endoperoxide. This secondary metabolite was obtained from the roots of Scleria striatinux
de Wild (syn. S. striatonux
) [Cyperaceae], a local spice in parts of Cameroon. The roots of S. striatinux
are also used as an herbal tea for fevers. This study was prompted by the observation that a CH2
/MeOH extract of S. striatinux
was moderately active against both chloroquine-sensitive and -resistant strains of Plasmodium falciparum
(cf., , entry 1).
Structure and relative configuration of okundoperoxide (1). The absolute configuration of 1 is undetermined; the structure of the enantiomer portrayed has been arbitrarily chosen.
Antiplasmodial Activity of S. striatinux: Crude Extract Versus Secondary Metabolite
A sample of S. striatinux
roots was harvested in Oku in the Northwest Province of Cameroon. Following gradient chromatography on silica gel and subsequent size exclusion chromatography (SEC) on Sephadex LH-20, the extract of the dried roots gave okundoperoxide (1
0.3 in 3:2 hexanes:EtOAc). Okundoperoxide was unstable to gas chromatographic analysis,3
an observation that became understandable once the peroxide subunit was identified. High resolution ESI-MS analysis gave a sodiated parent (and base peak) ion of mass 289.1402. The 13
C NMR spectrum in CDCl3
indicated the presence of 15 carbon atoms. Together, these data showed a molecular formula of C15
and five degrees of unsaturation.
Key 1H and 13C NMR data are reported in . All 15 carbon and 22 (first order) proton resonances were identified. The 13C NMR spectrum contained resonances for one ketone and four olefinic carbons. The 1H NMR spectrum suggested the presence of four methyl groups (one allylic with only long-range coupling and three aliphatic singlets) and three olefinic, one oxymethine, and one pair of oxymethylene protons. The HMQC spectrum clearly showed one-bond correlations that are the primary basis for the assignments of carbon chemical shifts listed in .
13C and 1H NMR Spectroscopic Data for Okundoperoxide (1) (CDCl3, 75 and 500 MHz)
The IR spectrum showed characteristic absorption bands for hydroxy (3477 cm-1) and carbonyl (1674 cm-1) groups. The former was consistent with a one-proton resonance at δ 1.36 ppm, which disappeared in a deuterium exchange experiment. The carbonyl absorption was suggestive of a conjugated enone, which was supported in the NMR spectrum by the chemical shifts of olefinic proton (δ 6.73 and 5.94) and carbon (δ 150.3 and 127.9) signals and of the carbonyl carbon resonance (δ 203.2). These data, together with the doublets of the olefinic proton resonances (J = 10.2 Hz), pointed to a 4,4-disubstituted Z-enone moiety.
The 1H-1H COSY spectrum indicated an isolated four spin system that included the proton at δ 1.96, having three large coupling constants (13.0, 13.0, 11.2 Hz). This was indicative of an axial-like methylene proton in a six-membered ring, flanked by two vicinal, trans methine protons (-CHCHaxHeqCH-). A COSY correlation between resonances for the olefinic proton at δ 5.75 and the methylene pair centered at δ 4.26 indicated a trisubstituted olefin bearing an oxymethylene group. The connectivity pattern deduced from the HMBC spectrum integrated the above subunits, along with the four methyl groups, into a common constitution. Specifically, structure 1 was consistent with all of the COSY and HMBC correlation data.
In addition to the 1,3-diaxial nature of H-4 and H-6 deduced from the coupling constant analysis, the remaining relative configurations shown in 1
were assigned largely on the basis of NOE observations (). The acyclic E
-olefin geometry is indicated by the enhancement of H-1 by H-15.4
Mutual enhancements of H-4 and H-6 reaffirm their cis-relationship. The trans
nature of the ring fusion was deduced from the sets of NOEs among H-5ax
/H-12/H-14 and H-4/H-5eq
Three-dimensional representation of 1 with key NOE correlations shown; these were used to assign the relative configurations among the three stereogenic Csp3-centers and the geometry of the acyclic alkene.
With the intent of reducing the peroxide bond in 1
with triphenylphosphine via an intermediate like 6
(), we treated a sample of 1
P in CDCl3
and monitored the subsequent events by 1
H NMR spectroscopy. Somewhat surprisingly, there was no observable change at ambient temperature. Moreover, when the reaction solution was heated in a 65 °C bath, the major product formed was the furan 2
, which has the same overall oxidation state as 1
and is the result of a net dehydration reaction. We suspect that enone 3
is an intermediate in this transformation. Zwitterion 6
, if formed, could preferentially undergo intramolecular elimination of phosphine (see arrows in 6
) rather than, for example, cyclization to a fused tetrahydrofuran derivative via displacement of triphenylphosphine oxide. Alternatively, the hindered nature of the dialkylperoxide in 1
may have induced a different reaction course from the outset. Namely, the phosphine may have functioned preferentially as a base rather than as a reductant to effect an eliminative opening via loss of H-4 and cleavage of the peroxide O-O bond (see arrows in 1
) to give the enone 3
-γ-Hydroxy-α,β-enones similar to 3
are known to undergo spontaneous isomerization and dehydration reactions to give furans.5
- to Z
-isomerization to convert 3
could involve a reversibly formed, rotatable intermediate epoxide (cf. 7a
) or triphenylphosphine adduct (cf
). There are many reported examples of dehydration of Z
-γ-hydroxy-α,β-enones like 4
under mild conditions to give the corresponding furans,7
likely via hemiketals like 5
. It is notable that among the many thermal decomposition products observed upon GC-MS analysis of okundoperoxide (1
), the furan 2
was the most abundant.3
Finally, the R- and the S-Mosher ester (methoxytrifluoromethylphenylacetyl, MTPA) derivatives of the alcohol 1 (8R and 8S, respectively, in ) were prepared using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) and the R- and S-Mosher acid (MTPA-OH), respectively. The 1H NMR data for these esters do not allow us to deduce the absolute configuration of 1 because of the large distance between the MTPA and substrate stereogenic centers. However, the spectra of these diastereomers are distinguishable, which should be helpful for later assignment of absolute configuration upon synthesis of one enantiomer of 1, an endeavor we are pursuing.
Figure 3 The two possible structures for each of 8R and 8S, the R- and S-Mosher ester derivatives of 1, prepared8 via EDC coupling of 1 with R- and S-MTPA-OH, respectively.
Antiplasmodial activity for the initial crude extract and for purified okundoperoxide (1) are shown in . Although the sample of 1 used for the biological testing was only 90% pure, we attribute the observed antiplasmodial activity to 1 because the interfering material, which was also isolated from a partially overlapped chromatographic fraction, is inactive in this assay. Finally, the presence of other compounds with antiplasmodial activity in S. striatinux cannot be ruled out. Further investigation of this possibility is also ongoing.