In vitro antiplasmodial activity of Dicoma anomala subsp. gerrardii (Asteraceae): identification of its main active constituent and gene expression profiling

To determine the transcriptional effects of a novel plant-based compound, dehydrobrachylaenolide, on P. falciparum, parasite cultures were treated with the compound over time. Samples were taken for analysis 2, 6, and 12 hours post-invasion of human red blood cells. Control cultures were treated simultaneously with DMSO, and samples isolated at 2, 6, and 12 hours for transcriptional analysis. Background Antimalarial drug resistance threatens to undermine efforts to eliminate this deadly disease. The resulting omnipresent requirement for drugs with novel modes of action prompted a national consortium initiative to discover new antiplasmodial agents from South African medicinal plants. One of the plants selected for investigation was Dicoma anomala subsp. gerrardii, based on its ethnomedicinal profile. Methods Standard phytochemical analysis techniques including solvent-solvent extraction, thin-layer and column chromatography, were used to isolate the main active constituent of Dicoma anomala subsp. gerrardii. The crystallised pure compound was identified using nuclear magnetic resonance spectroscopy, mass spectrometry and X-ray crystallography. The compound was tested in vitro on Plasmodium falciparum cultures using the parasite lactate dehydrogenase assay. The effects of treatment on the P. falciparum transcriptome were subsequently investigated by treating ring-stage parasites (alongside untreated controls) with the pure compound, followed by oligonucleotide microarray and data analysis. Results The main active constituent was identified as dehydrobrachylaenolide, a eudesmanolide-type sesquiterpene lactone. The compound demonstrated an in vitro IC50 of 245.6 nM, which was comparable to the IC50 of chloroquine, against a chloroquine-resistant strain (K1) of P. falciparum. Microarray data analysis identified a cluster of unique genes that were differentially expressed as a result of the treatment and gene ontology analysis identified various biological processes that were significantly affected. Comparison of the dehydrobrachylaenolide treatment transcriptional dataset with a published artesunate (also a sesquiterpene lactone) dataset revealed little overlap. This suggests differentiated modes of action between the two compounds. Conclusions Dehydrobrachylaenolide could play a valuable role as a drug candidate to generate new antimalarial compounds with novel modes of action and favourable ADMET properties. Reference design. Transcriptional analysis was performed as described in GSE18075 with the following modification. Profiles from parasite cultures isolated 2hrs after drug treatment during early ring-stage development were contrasted against untreated control cultures isolated at this timepoint. Profiles from parasites isolated 6 and 12 hours following drug treatment were contrasted to against untreated control cultures isolated at 6 hours. A detailed description of the statistical methodology used for this dataset is outlined in the accompanying manuscript. A reference design was employed for array hybridisation, utilising the URR pool described previously in the NCBI GEO Series GSE18075 dataset. All solvent-control and drug-treated samples were hybridised to Operon slides, along with the URR. In contrast to the hybridisation protocol described in GSE18075, 40 pmoles of each dye was hybridised to each array, utilising Cy3 (sample) and Cy5 (reference) dyes from GE Healthcare (#RPN5661), specifically optimised for nucleic acid labelling. A total of nineteen slides were processed in the study. Two independent cDNA samples (biological replicates) were prepared for each untreated and drug-treated sample at each time point. One of the biological replicate cDNA samples were additionally hybridised to a third slide (representing the technical replicate). In the case of the 2 hour DMSO treated control culture, an additional technical microarray replicate was included for quality control purposes. GenePix results (gpr) files were generated using GenePix 6.0 (Molecular Devices) software, without normalization. For clustering analyses, results files were normalized with DNMAD (Diagnosis and Normalization for MicroArray Data) using print-tip loess. The normalized values were subsequently downloaded and analyzed with the Multiexperiment Viewer (MeV) in the TM4 software suite. Hierarchical Clustering (HCL, average linkage) was performed to estimate technical and biological variation between samples and at which point cytostasis most likely occurred for comparative purposes in downstream analyses. Intensity data for individual slides were imported into LIMMA (linear models for microarray data) in the R computing environment. Pre- and post-normalization diagnostic plots were performed using MARRAY. Data from each microarray slide was normalized using print-tip loess. Data between microarrays was normalized using R-Quantile normalisation. Pearson correlations were computed in MS Excel to estimate variation between technical and biological replicates. Spots excluded from slide correlations and normalisation were those weighted by the limma script or flagged in the genepix results file (gpr). Additionally, spots termed Alien, Empty, Null and Operon Use Only were excluded from the correlation analyses. These spots were similarly excluded for correlations between untreated and treated samples at each time point following normalisation. Results from biological and slide replicates within each of the time points were collated, and linear models were computed to contrast gene expression between time points. A two-fold change in gene expression was used as cut-off, in conjunction with correction for false discovery (false discovery rate (FDR) = 5%). Analysis of differentially expressed genes was performed in MADIBA (Micro Array Data Interface for Biological Annotation).