Plasmodium falciparum malaria is responsible for nearly half a million deaths every year, despite the recent worldwide reduction in the incidence of the disease. Worryingly, the trend of decreasing malaria mortality has stagnated since 2015, followed recently by a steep increase due to the impact of the Covid-19 pandemic (WHO, 2021). This later tendency highlights the necessity to develop new strategies to tackle the disease, which in turn requires new valuable knowledge on parasite biology.

Plasmodium falciparum is notorious for its high capacity to develop drug resistance. Since the first reports of quinine resistance, more than 100 years ago (Rodrigues Coura, 1987), this parasite has demonstrated the ability to circumvent the action of essentially every antimalarial treatment deployed to a significant scale. Artemisinin combination therapies (ACTs) were introduced to curb the phenomenon of drug resistance, being the mainstay in uncomplicated malaria therapy since their global adoption during the first decade of the 21st century. Nevertheless, even for these generally highly efficacious therapies, reports suggest a progressive trend of reduction in efficacy. This has been first recognized for artesunate-mefloquine (Price et al., 2004) and artesunate-amodiaquine (Holmgren et al., 2006; Humphreys et al., 2007). In addition, the recent failure of dihydroartemisinin-piperaquine (DHA-PPQ) therapy in Cambodia just a few years after its formal implementation in the country, showcases the remarkable resilience of this parasite to drug pressure (Amaratunga et al., 2016; Amato et al., 2017; Hamilton et al., 2019). Nevertheless, and until recently, the globally most used ACT, artemether-lumefantrine (AL) has shown remarkable resilience.

In Africa, the global malaria epicenter, AL is the key ACT, present in essentially every national malaria program of the continent and represents the backbone of malaria clinical management. Accordingly, recent reports of decreased AL efficacy have been met with considerable concern (Plucinski et al., 2015; Plucinski et al., 2017).

The main culprit of artemisinin reduced susceptibility, clinically characterized as an infection with a clearance half-time of >5 h (Dondorp et al., 2009), has been identified as the P. falciparum Kelch 13 (PfK13) (Straimer et al., 2015), with other players possibly augmenting the susceptibility (Veiga et al., 2014). However, therapy failure has been more likely to result from failure of the long half-life partner drug. As so, it is of importance to monitor and find novel markers of partner drug resistance. At present, most partner drugs are from the quinoline family and while a complete understanding of the molecular basis of resistance is still not clarified, two transporters, P. falciparum multidrug resistance protein 1 (PfMDR1) and P. falciparum chloroquine resistance transporter (PfCRT), have been shown to be pivotal in the parasite response to ACT partner drugs.

PfMDR1 (also referred to as the P-glycoprotein homologue, Pgh) is a ca. 1419 amino acid protein comprising 12 trans-membrane domains, and a member of the ATP-binding cassette (ABC) superfamily (Foote et al., 1989). PfMDR1 is oriented toward the digestive vacuole (DV) lumen and works as an importer of solutes toward this organelle, including antimalarials (Rohrbach et al., 2006). PfMDR1 single nucleotide polymorphisms (SNPs), and of particular importance, the N86Y allele, have been associated with in vivo and in vitro sensitivity to a range of quinoline antimalarials (e.g., mefloquine, lumefantrine, amodiaquine) and artemisinin derivatives (Veiga et al., 2016; Gil and Krishna, 2017).

PfCRT was identified as the main determinant of chloroquine resistance (Fidock et al., 2000). PfCRT is a 424 amino acid protein with 10 transmembrane domains, a member of the drug-metabolite transport superfamily (Martin and Kirk, 2004), and localizes in the DV membrane of the parasite being able to pump antimalarial drugs out of this organelle (Martin et al., 2009). The PfCRT physiological function has been proposed to be related to the export of host-derived peptides to the cytoplasm (Shafik et al., 2020). The critical chloroquine resistance mutation K76T and other polymorphisms have been documented to influence parasite sensitivity to quinolines (Sidhu et al., 2002; Johnson et al., 2004; Sisowath et al., 2009; Agrawal et al., 2017) and artemisinin compounds (Sidhu et al., 2002; Cooper et al., 2005).

Increased gene expression has been associated with drug resistance in Plasmodium falciparum by the causal link of augmented protein levels and treatment outcome. This is evidenced by pfmdr1 duplications leading to mefloquine and lumefantrine decreased efficacy (Price et al., 2006; Calcada et al., 2020), and through the association of plasmepsin 2 and 3 amplifications with the collapse of piperaquine containing ACTs in Southeast Asia (Silva et al., 2020). From this background, it is conceivable that changes in transcription activity can also play a role in therapy response. Accordingly, we have previously shown that the levels of pfk13 expression negatively correlate with P. falciparum infection clearance time (Silva et al., 2019) upon AL treatment. In this study, we sought to evaluate the in vivo gene expression of the two most-known modulators of drug resistance, pfcrt and pfmdr1, during AL treatment.

Protein levels associated with increased gene expression have been related to several drug resistant-phenotypes in P. falciparum (Price et al., 2004; Sidhu et al., 2006; Sharma et al., 2013; Amato et al., 2017; Inoue et al., 2018; Calcada et al., 2020).

PfMDR1 expression has been implicated in the resistance to multiple drugs, mostly because of an increased gene copy number. Of note, increased PfMDR1 expression is associated with mefloquine resistance and can lead to the inefficacy of the artesunate-mefloquine combination (Price et al., 2004). Gene expression can modulate protein levels and is a potential mechanism to modulate drug response. There are very few studies assessing the patterns of P. falciparum gene expression on ACT in uncomplicated malaria patients. Of interest, albeit artemisinin resistance is mainly associated with decreased PfK13 levels resultant of SNPs (Birnbaum et al., 2020), lower pfk13 transcript levels in vivo have also been linked with longer parasite clearance times, along the course of AL treatment (Silva et al., 2019). In this work, we focused on two key players of parasite resistance against the long half-life quinoline partner drugs, the pfmdr1 and pfcrt.

A large range of baseline expression levels was observed for both pfcrt and pfmdr1, of approximately 10 and 30-fold, respectively. For pfmdr1, these in vivo data are supportive of previous expression studies using clinical isolates, where a similar range was reported (Legrand et al., 2012). The reasons for these inter-infection differences are unclear. The possibility of a significant influence from the presence of other intra-erythrocytic stages should be relatively small, considering that it involved patients experiencing fever, a status known to drive parasite cycle synchronization (Kwiatkowski, 1989; Engelbrecht and Coetzer, 2013). Moreover, even if there would be an influence of later stages inter-stage, differences in expression in vitro are not in the same order of magnitude, compared with those observed in our study (Young et al., 2005; Veiga et al., 2010).

A central observation was the clear correlation between the expression of both pfcrt and pfmdr1. This was most evident before treatment initiation but extended itself to post AL drug exposure. These data support in vitro reports by Adjalley et al. (2015) with individual clones, which further linked these transporters inside a larger set of co-regulated loci. Our work shows that this is a regular event during infection, supporting a complementary role of these food vacuole-located transporters in the parasite regular physiology.

Albeit the majority of the studied parasites did not experience an induction on their pfmdr1 and pfcrt genes, our work shows that such events are not apparently uncommon during infections (particularly pfmdr1) (Figure 2). These observations raise the issue of how much “hard-wired” the parasite transcription is when considering clinically relevant in vivo situations and analyzing a sufficiently large sample. For most in vitro studies—including our own (Veiga et al., 2010)—drug exposure has not been reported to drive large transcriptomic changes. Large differences in the baseline gene expression between P. falciparum strains have been proposed to be slow and likely driven by epigenetic processes (Batugedara et al., 2017; Arama et al., 2018). Such processes have been, to a certain extent, showcased in the in vitro selection of blasticidin-resistant parasites, based on the epigenetic driven suppression of clag3 gene expression (Sharma et al., 2013; Chan et al., 2020). From this perspective, it is possible that what we are observing in vivo are minor sub-populations already expressing pfmdr1 and pfcrt at more than average levels that happen to be selected during drug treatment. Epigenetic-based mechanisms might co-exist with events of promoter-driven gene upregulation. As referred, increases of >2-fold have been reported in vitro upon exposure to common antimalarials using clonal populations (Veiga et al., 2010).

The molecular basis of the observed pfmdr1/pfcrt co-expression is not known. Of note, both genes harbor common putative xenobiotic response DNA elements in their 5’ promoter regions (up to 5 Kb upstream of the transcription initiation), especially nuclear receptor-recognition sequences (Johnson et al., 2008). Of interest for this discussion, a number of those are common for both pfcrt and pfmdr1, including BARBIE (barbiturate-inducible element) boxes, retinoic acid receptor (RAR) and retinoic acid receptor half-site (RXR) elements, and progesterone receptor-binding sites (PRE). It is conceivable that, similar to other eukaryotic systems (Penvose et al., 2019), some of these elements might be mechanistically associated with the observed co-regulation pattern between these two genes. In fact, artemisinin derivatives are known agonist ligands of the nuclear pregnane-X-receptor (PXR) and constitutive androstane receptor (CAR) in higher mammals, leading to an increased activity of targeted genes, through the interaction with regulatory sequences as previously mentioned (Burk et al., 2005; Zang et al., 2014). This includes genes coding for P-glycoprotein (Pgp)-type ABC transporters, evolutionary related to the parasite pfmdr1 product. Moreover, investigations using luciferase expression constructs to probe pfmdr1 promoter regions have shown these as significantly responsive (Myrick et al., 2003). Furthermore, when applying the classical ADME inducer phenobarbital, five- to sixfold increases in the PfMDR1 protein content were also reported, albeit PfCRT stayed non-responsive, suggesting no significant involvement in our observations of the BARBIE Box (Johnson et al., 2008). Moreover, it is conceivable that the opportunity for pfmdr1 to be induced might be ruled by the epigenetic status of the proximal 5′ promoter, with favorable conditions only statistically happening in a fraction of the parasite population at any time point. Further studies will be required to investigate this possibility.

An increased copy number of N86 carrying PfMDR1, which leads to significant increases in this gene transcript and associated protein, is a well-documented factor in the in vivo and in vitro parasite susceptibility to aminoalcohol quinolines and artemisinins (Veiga et al., 2011; Calcada et al., 2020; Silva et al., 2020). Linking this information with the known importance of the PfMDR1 N86 allele modulating AL therapy in Africa (Malmberg et al., 2013; Venkatesan et al., 2014), we initially hypothesized that in a clinical setting, the carriers of this critical allele would tend to have significantly higher baseline levels of expression. Our results do not support this hypothesis, as N86 carrying parasites did not show a significant difference in pfmdr1 expression compared with the Y86 ones. In fact, if we consider any trend, it is for a higher expression among the 86Y carriers. The consistently observed selection of 86N alleles during AL treatment is more likely to be associated with SNP-driven changes in the structure of the protein, as predicted from in silico studies (Ferreira et al., 2011) and not through a link with increased transcriptional activity.

A major limitation of the present study is the participant numbers, a factor related to the outstanding demands of these types of field studies. In addition, the nature of the study limited the availability of biological material for analysis, precluding the consistent use of multiple internal qPCR controls. However, the project’s exploratory nature did not prevent us from providing new insights into in vivo gene expression of two key markers of antimalarial resistance.

In conclusion, this work represents, to our knowledge, the first detailed qPCR-based analysis of the parasite pfcrt and pfmdr1 genes during an ACT clinical exposure. By challenging the general view of a rather inert parasite transcriptome, at least concerning the genes under focus, we highlight the need to link valuable in vitro data on the parasite drug response mechanisms to the in vivo therapy context.