Although PN is well-recognized as a highly potent anti-malarial drug against both sensitive and multi-drug resistant lines of malaria parasites, its molecular mechanism of action is not fully understood. Previous studies have demonstrated that its primary mode of action is the same as CQ, namely interaction with haem and subsequent inhibition of β-haematin assembly. Given that both drugs are structurally related and are proposed to act on the same target(s), it was hypothesized that they may induce similar transcriptomic changes at IC50 concentrations.
The transcriptome of chloroquine-resistant P. falciparum K1 strain in response to PN and CQ exposure was investigated. Similar to previous microarray studies of P. falciparum under anti-malarial drug treatment [36, 42], the magnitude of change in expression for most genes in response to CQ and PN was low. Nonetheless, 297 features showed significant change to PN and 205 to CQ. The overlap of 40 features among the PN and CQ features was highly significant, suggestive of a similar genetic response. In contrast, the PN and CQ features overlapping with those responsive to an unrelated drug, Pyrimethamine in the same K1 parasite strain exposed for 24 hours were not significant. In total, 461 features showed significant changes in response to either or both PN and CQ drugs. A major difficulty in interpreting these data with respect to primary mechanism of drug action is confounding responses not directly related to the mechanism of drug action. Although no delay of parasite development was observed in CQ and PN drug-treated cultures by comparison of morphology, it is known that transcriptional arrest precedes morphological arrest . It should be noted that the term transcriptional arrest relates to a delay in the stage-specific pattern of gene expression within a short time-frame, i.e. the same 48 h developmental cycle. This phenomenon is probably quite distinct from the changes occurring in recrudescing parasites in the longer time-scale of a clinical treatment period (3-4 days). Such a delay in the developmental programme of gene expression may, therefore, have a confounding effect on identification of drug-specific responders. A conservative approach was taken to address this confounding effect, in which all stage-specific transcripts from the corresponding developmental stages during the different drug treatment periods were removed. This approach, however, could lead to removal of genes which are primary drug responsive, yet developmentally regulated.
Besides the primary effects of PN and CQ treatment, i.e. inhibition of α-haematin formation and glutathione-dependent haem degradation, secondary/downstream effects of these drugs on the parasite have been described. At very high concentrations, CQ and other quinoline drugs disrupt vesicular transport [43, 44]. Moreover, CQ inhibits translation , disrupts mitochondrial membrane potential and induces DNA fragmentation . The PN and CQ responsive microarray features with gene annotations matching the perturbation dataset  could be considered secondary responses. The gene annotations encompass a variety of cellular functions, although none are significantly associated with the aforementioned CQ secondary effects. Microarray features corresponding to the Pfcrt gene, which mediates resistance to CQ, showed significant changes in expression in response to PN and CQ. However, the Pfcrt gene is present in the perturbation dataset and cannot be considered as a primary PN or CQ responsive gene. Moreover, no significant changes in expression were found for features corresponding to other CQ and quinoline drug transporter genes reported in the literature (Pfmdr1, Pfmdr 2, Pfmrp1, Pfmrp2 and Pfnhe1).
After filtering by comparison with the perturbation dataset and removal of genes known to be up-regulated in gametogenesis, the remaining 46 PN and CQ responsive features showed a significant over-representation of genes encoding host-exported proteins. These genes belong to large gene families, including var, stevor, surfin, rifin and phist. A previous study of the transcriptional response of P. falciparum under IC50 CQ treatment showed that 36 genes could be considered as primary CQ responsive, with many more genes showing changes in expression because of differences in staging and secondary effects . Of the 36 primary CQ genes reported in , two rifin genes (PF10_0006 and PF10_0404) are also present among the annotated genes represented by the 46 PN/CQ responsive features after filtering. Although the data are suggestive of primary PN and CQ responsive genes, other perturbations such as antifolates , and the bisthiazolium compound T4  also induce changes in expression of host-exported proteins of multi-gene families, albeit for separate genes. Despite some common features of transcriptional regulation, including epigenetic switching mechanisms and sharing of transcriptional activators , very little is known of how the expressions of these gene families are modulated by environmental perturbation. Given that different drugs exert responses in these gene families, their roles as potential mediators of drug resistance should be investigated further.
Comparison of the drug-responsive genes shows that PN appears to exert a faster-acting transcriptional perturbation than CQ. From the four drug treatment conditions tested in this study, the PN 4 h treatment was globally more similar to the CQ 24 h treatment than CQ 4 h. Similarities between the PN and CQ transcriptional responses were also noted in overlapping genes and functional categories, which support the hypothesis that the genes responsive to the two drugs are similar. It should be noted that because PN is far more potent than CQ against the K1 strain, much higher concentrations of the latter drug were needed to produce similar biological effects. Therefore, the differences in cytosolic concentrations between PN and CQ may contribute to some of the minor observed differences between the PN and CQ transcriptional responses. A simple explanation for why PN is faster acting and thus more effective than CQ, especially against CQ-resistant parasites, can be made based on the known chemical properties. PN is more hydrophobic than CQ, with a log P value comparable to amodiaquine [49, 50] and thus the uptake into the food vacuole, the site of action, will be faster for PN. Furthermore, PN has an extra highly basic group, which means it may be protonated more readily and thus accumulate more rapidly in the food vacuole than CQ.