P. falciparum artemisinin resistance: the effect of heme, protein damage, and parasite cell stress response

Abstract

Despite a significant decline in morbidity and mortality over the last two decades, in 2018 there were 228 million reported cases of malaria and 405,000 malaria-related deaths. Artemisinin, the cornerstone of artemisinin-based combination therapies, is the most potent drug in the antimalarial armamentarium against falciparum malaria. Heme-mediated activation of artemisinin and its derivatives results in widespread parasite protein alkylation, which is thought to lead to parasite death. Alarmingly, cases of decreased artemisinin efficacy have been widely detected across Cambodia and in neighboring countries, and a few cases have been reported in the Guiana Shield, India, and Africa. The grim prospect of widespread artemisinin resistance propelled a concerted effort to understand the mechanisms of artemisinin action and resistance. Identification of genetic markers and knowledge of molecular mechanisms underpinning artemisinin resistance allow prospective surveillance and inform future drug development strategies, respectively. Here, we highlight recent advances in our understanding of how parasite vesicle trafficking, hemoglobin digestion, and cell stress responses contribute to artemisinin resistance.

Graphical Abstract

Heme-mediated activation of artemisinins causes widespread damage to Plasmodium parasites. Though the exact mechanism of resistance has not been fully elucidated, resistance may be mediated by a decrease in available free heme and/or an increased parasite stress response capacity. Here, we review the contribution of parasite genetics to artemisinin resistance and how these genes, including K13, may mediate artemisinin resistance.

Artemisinins: potent pro-drugs with specificity for heme-digesting parasites

Malaria remains a global health disease, with 228 million reported cases and 405,000 deaths documented in 2018 (1). The World Health Organization recommends artemisinin-based combination therapies (ACTs) for the treatment of uncomplicated falciparum malaria, which is defined as Plasmodium falciparum infection diagnosed by microscopy or a rapid diagnostic test, and the absence of symptoms associated with severe malaria such as convulsions, severe anemia, hypoglycemia, impaired consciousness, impairment of organs (lungs, liver, kidneys), or hyperparasitemia (>10% parasitemia i.e. >10 parasites per 100 red blood cells, RBC) (1). ACTs are also used for treatment of unconfirmed Plasmodium species in regions of chloroquine resistance (1). The three-day ACT regimens are comprised of a fast-acting but short-lived artemisinin derivative (Figure 1A) paired with a longer-lived partner drug. Depending on geographical parasite resistance profiles, one of six ACTs is recommended: artemether and lumefantrine (AM-LF), artesunate and amodiaquine (AS-ADQ), artesunate and mefloquine (AS-MF), dihydroartemisinin and piperaquine (DHA-PQP), artesunate and sulfadoxine-pyrimethamine (AS-SP), or artesunate-pyronaridine (AS-PYR) (1). The semi-synthetic artemisinin derivatives AM and AS are converted in vivo to DHA. Artemisinin and its derivatives are active against asexual blood stages as well as male but not female gametocytes (2). Artemisinin is a potent antimalarial that inhibits parasite-mediated heme detoxification (3, 4), generates intra-parasite reactive oxygen species (ROS) (5, 6), and non-specifically alkylates parasite proteins (7, 8, 9). These protein aggregates overwhelm the parasite cellular quality control system (10) and are thought to lead to parasite death.

Figure 1.

(A) Structure of artemisinin and its derivatives. The endoperoxide bridge, indicated in red, is required for antimalarial activity. (B) Intraerythrocytic developmental stages of P. falciparum. Representative images of parasites at various hours post invasion stained with Giemsa and visualized with a light microscope are shown. Hemoglobin uptake and digestion increases as parasites progress from rings to schizonts, as indicated by a deeper shade of red. (C) Activation of dihydroartemisinin. Cleavage of the endoperoxide bridge is initiated by heme (Fe²⁺-FPIX), forming a short-lived alkoxy radical. Thermodynamic induced rearrangement produces a primary carbon-centered free radical, indicated in red, that then alkylates heme and parasite proteins.

The complex life cycle of malaria parasites both increases challenges of elimination and provides unique opportunities for intervention. Within minutes of transmission from an Anopheles mosquito bite, Plasmodium parasites home in to hepatocytes where they replicate asexually within a parasitophorous vacuole before being released into the bloodstream. Here they invade erythrocytes and enter the asexual blood stages, which are responsible for clinical symptoms. Up to 80% of host hemoglobin is imported into parasites (11). Hemoglobin uptake begins in very early rings, likely by micropinocytosis (12). In trophozoite and schizont stage parasites, hemoglobin is ingested through endocytosis via cytostomes, which are double membrane invaginations of the parasite plasma membrane and parasitophorous vacuolar membrane (12). As parasites progress from ring to trophozoite stages, artemisinin sensitivity increases, mirroring an increase in hemoglobin uptake and digestion, with the exception of very early rings (2–4 hours post-invasion, hpi) that demonstrate hypersensitivity to artemisinin (Figure 1B) (13). In the parasite digestive vacuole (DV), hemoglobin is proteolytically cleaved by aspartic and cysteine proteases (12). Hemoglobin digestion releases redox-active free heme (ferroprotoporphyrin IX, Fe²⁺-FPIX), which is oxidized to hemin (α-hematin, or ferriprotoporphyrin IX, Fe³⁺-FPIX); both are toxic to the parasite. Therefore, parasites convert α-hematin to inert β-hematin, which then undergoes biocrystallization to form hemozoin or “malaria pigment” (11). Yet, there is enough free heme to activate artemisinin (14), leading to exquisite potency against hemoglobin-digesting Plasmodium parasites (15).

Whether artemisinin is activated by heme or ferrous iron in vivo has been a long-standing question. Several lines of evidence point to the fact that biologically, heme is the primary activator of artemisinin and related endoperoxides. Biochemically, heme activates artemisinin at a faster rate compared to inorganic ferrous iron (Fe²⁺) (16). Whereas the cysteine protease inhibitor ALLN abrogated the ability of an artemisinin-based activity probe to alkylate parasite proteins (7), addition of chelators showed no effect on probe binding (7, 8). In addition, inhibition of hemoglobin digestion by genetic deletion of the cysteine proteases falcipain-2 and falcipain-3 or chemical inhibition using the cysteine protease inhibitor E64d lead to a marked 80–100% increase in parasite survival to artemisinin; in comparison, only at most ~30% rescue was seen when iron chelators were employed (17). Cleavage of the endoperoxide bond is key to the antimalarial activity of artemisinin and related structures (18). This break is initiated by an electron transfer from free heme, forming a short-lived alkoxy radical. β-scission rearrangement produces primary carbon-centered free radicals that alkylate nearby heme and proteins (18), generating toxicity to the parasite (Figure 1C).

To date, the exact mechanism of action of artemisinin and related endoperoxides remains unsolved. Artemisinin-mediated parasite killing involves peroxide bond-dependent (15) alkylating activity (19). Chemical proteomics exploiting artemisinin activity-based probes coupled to click chemistry followed by mass spectrometry analyses was performed by three separate groups to identify parasite proteins alkylated by artemisinin (7, 8, 9). These studies identified alkylation of parasite proteins involved in numerous processes including hemoglobin digestion, metabolic processes, and antioxidant defenses. Due to the breadth of parasite protein alkylation, it is hypothesized that artemisinin-mediated protein alkylation results in death by functionally inhibiting key proteins involved in essential parasite pathways. Note that there was little correlation in identified alkylated proteins between the three studies, which had differences in approach and controls (7, 8, 9), suggesting that activated artemisinin promiscuously and non-specifically alkylates proteins within a proximal vicinity.

Artemisinin-mediated heme alkylation has been shown to occur both in vitro (20) and in vivo (19). Heme-artemisinin adducts block conversion of α-hematin to hemozoin even more potently than chloroquine (4), a well-established inhibitor of hemozoin formation (3). In addition, the heme in these monomeric adducts is unable to be converted to hemozoin due to stearic hindrance, and thus may retain free heme’s toxicity and redox potential (4). Artemisinin potently kills P. falciparum by inhibiting key parasite processes. Nevertheless, drug pressure has driven the evolution of artemisinin-resistant parasites. The contribution of parasite processes in artemisinin survival responses are examined later in the review.

Artemisinin resistance

Clinically, artemisinin resistance is characterized by parasites that display clearance times greater than three days or parasite clearance half-lives of greater than five hours after patients are treated with artemisinin or ACT (21, 22). First detected in Western Cambodia (21, 22), resistance has now spread to four neighboring countries: Thailand (23, 24, 25), Laos (24), Vietnam (24, 26), and Myanmar (24, 27). A few cases of artemisinin resistance outside of Southeast Asia have been reported in Equatorial Guinea (28), Uganda (29), Senegal (30), India (31), and the Guiana Shield (32, 33). Artemisinin resistance is not yet widely detected in Africa, where 93% of cases and 94% of all malaria deaths occur (1). Clinically, parasite clearance times depend on a number of factors, including the host immune system, drug quality, adherence to medication schedule, and innate parasite characteristics. Here, we will focus on the contribution of parasite genetics in mediating delayed parasite clearance times.

Due to the peculiarities of artemisinin resistance, standard 72 hour in vitro growth inhibition assays measuring half-maximal inhibitory concentration (IC₅₀) of asynchronous cultures are unable to robustly distinguish between fast- and slow-clearing infections (22). Instead, the ring-stage survival assay (RSA_0–3h, also referred to as RSA) is required to delineate artemisinin sensitivity (34). Early ring-stage (0–3 hpi) parasites are exposed to a pharmacologically-relevant dose (700 nM) of DHA for 6 hours, then parasite viability is assessed 66 hours later in the following replication cycle (34). A survival rate of greater than 1% in RSAs is a commonly-accepted threshold to classify parasites as artemisinin resistant (35, 36). Note that experimental variation can result in RSA values of 1–2% even for artemisinin-sensitive parasites, and thus RSA values should be interpreted with care. Inter-laboratory variations in conducting RSAs such as the length of DHA treatment (3 to 6 hours) and assessment of viability (plate reader vs. flow cytometry, staining of DNA only or inclusion of a marker of parasite viability) lead to slight variations in survival rates measured in vitro. Nonetheless, data from different groups are generally comparable.

Kelch 13: a molecular marker for artemisinin resistance

The breakthrough in understanding artemisinin resistance came through in vitro generation of the P. falciparum artemisinin-resistant strain (F32-ART5) (37) followed by whole genome sequencing, revealing mutations in seven genes (35). Cambodian parasites previously phenotyped for slow-clearing vs fast-clearing infections were examined for variations in these genes, revealing a strong correlation between mutations in the propeller domain of Kelch13 (K13, PF3D7_1343700) and increased parasite clearance times (35). The remaining six genes had wild-type sequence or intrapopulation variation and no correlation with survival rates in RSAs (35).

Kelch 13 is so named because it has kelch domains and is encoded on chromosome 13; it is not named based on homology to human KLHL13. The K13 gene is comprised of a Plasmodium-specific N-terminal domain followed by a CCC (coiled-coil-containing) domain, a BTB/POZ (Broad complex Tramtrack Bric-a-brack/Pox virus Zinc finger) domain, and a Kelch propeller domain. The six-bladed β-propeller domain consists of six Kelch motifs, each folding into a four stranded antiparallel β-sheet (38). Single nucleotide polymorphisms (SNPs) have been identified throughout the K13 gene, but only non-synonymous mutations in the propeller domain (K13^PDmut, K13 propeller domain mutant) are associated with delayed parasite clearance (24).

The particular K13 mutation (M476I) identified in F32-ART5 was not found in the slow-clearing Cambodian field isolates analyzed (35). Instead, these isolates had polymorphisms in K13 at Y493H, R539T, I543T, and C580Y (35); all five of these mutations lie in the propeller domains (35, 38). Parasite isolates possessing K13^M476I remain clinically sensitive to artemisinin (24, 39) and this is an exception to the rule that most K13^PDmut parasites display artemisinin resistance (24). Roughly 70% of Cambodian parasites harbor K13^PDmut (40), with the most common haplotype being K13^C580Y (41). Cambodia’s high prevalence of K13^PDmut (40) underlie the high rates of ACT treatment failure in this region (1).

The aforementioned SNPs in the K13^PD identified by Ariey et al. were individually confirmed to confer artemisinin resistance by K13 gene editing via clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) (42) or zinc finger nuclease (ZFN) (36). The availability of isogenic lines harboring K13^WT and K13^PDmut (M476I, Y493H, R539T, I543T, and C580Y) on distinct parasite genetic backgrounds have allowed different groups to independently investigate the contributions of K13^PDmut to artemisinin resistance (9, 10, 43, 44, 45, 46, 47, 48, 49).

K13: an adaptor for Cullin3-RING-E3 ubiquitin ligases (CRL3)?

K13 has low sequence homology (< 30%) to human Kelch proteins, which muddles assignment of function by homology. The crystal structure of K13 in the Protein Data Bank (https://www.rcsb.org/structure/4yy8) confirms earlier predictions of K13 propeller domain structure (35), and we eagerly await the publication of these efforts.

Multiple theories have been proposed as to how K13 mediates artemisinin resistance. A unifying thread is that K13 could act as an adaptor to bring E3 ligase(s) and its substrate(s) into proximity. Ubiquitin moieties are attached to substrates via an ATP-dependent cascade of E1 ubiquitin activating enzymes, E2 ubiquitin conjugating enzymes, and E3 ubiquitin ligases (50). K13 is hypothesized to bind a Cullin 3 RING E3 ligase complex (CRL3) via the BTB domain and bind a yet-to-be-identified substrate(s) via the kelch propeller domains (38, 51). In this proposed scenario, wild-type K13 (K13^WT) binds substrate(s) and allows for CRL3-mediated ubiquitination. In contrast, K13^PDmut would be unable to bind substrate(s), preventing CRL3-mediated ubiquitination (Figure 2). Depending on the nature of ubiquitin chain linkages, ubiquitination of substrate(s) can act as a signal for localization, downstream signaling, or proteasome-mediated degradation (50). Whether there are one or more substrates, discovering substrate(s) identity, and understanding how these substrates mediate survival in the presence of artemisinin is an area of active investigation. Several roles for K13 and pathways involved in artemisinin resistance have been proposed, outlined below.

Figure 2.

Potential role of K13 as an E3 substrate adaptor. A primed Cullin 3 RING E3 ligase complex (CRL3) is composed of Cullin 3 RING E3 ligase (CUL3), rbx1 (RING box protein 1), NEDD8, and E2 ubiquitin conjugating enzyme bound to ubiquitin. (A) K13^WT binds CRL3 via the BTB domain and substrate(s) via the kelch propeller domains, indicated as six curved lines arranged in a circle. Substrate(s) is/are ubiquitinated and could be targeted for proteasome-mediated degradation. (B) Artemisinin resistance-conferring mutations in the K13 propeller domain (K13^PDmut), indicated in green, prevent binding to substrate(s), sparing the substrate from subsequent degradation.

Vesicular trafficking

DHA inhibits P. falciparum phosphoinositide 3-kinase (PfPI3K, PF3D7_0515300) biochemically (52) and production of phosphatidylinositol 3-phosphate (PI3P) in P. falciparum-infected RBC (53). Increased levels of PfPI3K are detected in 3D7 and NF54 strains expressing the artemisinin resistance-conferring mutation K13^C580Y, but decreased amounts of PfPI3K co-precipitated with K13^C580Y (53), indicating that K13^C580Y is unable to recognize PfPI3K and target it for degradation. Indeed, K13^C580Y parasites were found to produce a greater number of PI3P vesicles compared to K13^WT parasites (54). With some exceptions, artemisinin resistance correlated with increased PI3P levels (53), though the exact involvement of PI3P in artemisinin resistance remains to be elucidated. Are K13 and PI3P involved in hemoglobin uptake?

K13 subcellular localization examined using GFP-tagged K13 or antibodies against K13 reveal that K13 localizes to cytostomes (47), is contained within vesicles (54), and appears in distinct punctate structures that localize to the food vacuole (55, 56). Of note, both K13^WT and K13^C580Y co-localize tightly (55). Using a novel dimerization-induced quantitative proximity-dependent biotin identification (DiQ-BioID) assay, K13 was shown to interact with the endocytic proteins Eps15-like protein (PF3D7_1025000) and AP2μ (PF3D7_1218300), but not clathrin heavy chain or light chain (PF3D7_1219100, PF3D7_1435500) (55), suggesting that K13 participates in non-clathrin-mediated endocytosis. Data supporting this unique mode of endocytosis in P. falciparum was shown in a separate mass spectrometry-based study that demonstrated the AP2 complex does not associate with clathrin heavy chain or light chain (56). This study identified kelch 10 but not K13 as an AP2μ-interacting partner (56).

Parasites with K13^PDmut (45) or mis-localized K13 (47) have decreased abundance of hemoglobin-derived peptides and metabolites (45, 47) as well as significantly reduced endocytosis at 3–6 hpi as measured by reduced uptake of fluorescent dextrans (55). Together, the data suggests that K13 may be involved in endocytosis of hemoglobin, and that this process is reduced in parasites with K13^PDmut or with reduced levels of K13^WT. Inhibition of hemoglobin endocytosis could be advantageous for artemisinin-treated parasites by limiting the amount of free heme available to activate artemisinin, which is consistent with findings of lower amounts of heme-DHA adducts in artemisinin-resistant P. falciparum K13^PDmut compared to K13^WT (57).

Oxidative stress

Free heme is redox-active, and parasites have a robust antioxidant system to handle ROS generated in the course of its normal lifecycle, as discussed in (58). Exposure to artemisinin further produces ROS (5, 6); do enhanced oxidative stress responses play a role in artemisinin resistance? In humans, oxidative stress prevents Kelch ECH associating protein 1 (Keap1) from facilitating proteasome-mediated degradation of nuclear factor erythroid 2-related factor (Nrf2), a transcription factor that upregulates genes encoding antioxidants (59). The propeller domain of K13 possesses 28% sequence homology to Keap1 and has been proposed to play an analogous role (35). In this scenario, K13^PDmut would be unable to bind and facilitate proteasome-mediated degradation of a Nrf2-like transcription factor. Increased basal antioxidant response levels that counter the onslaught of artemisinin-generated ROS would allow increased survival of K13^PDmut parasites. Although no Nrf2 ortholog has been discovered in the Plasmodium genome, there is evidence demonstrating the importance of antioxidants in contributing to artemisinin resistance.

In vitro-generated DHA-resistant parasite strains DHA1 and DHA2 demonstrated increased survival when pulsed with DHA at the ring-, trophozoite-, or schizont-stages (60), indicating there is a mechanism of artemisinin survival throughout the intraerythrocytic developmental cycle and that resistance is not confined to the early ring stages as previously thought (37). Gene analyses demonstrate increased expression of over 20 genes involved in antioxidant responses in DHA1 and DHA2 compared to the parental Dd2 strain, including superoxide dismutase (PF3D7_0623500), gamma-glutamylcysteine synthetase (PF3D7_0918900), thioredoxin-like protein 2 (PF3D7_0925500), thioredoxin peroxidase 2 (PF3D7_1215000), and seven heat shock proteins including heat shock protein 70 (hsp70, PF3D7_0818900) (60). Acute exposure to 1 μM DHA for two hours resulted in further upregulation of genes in the antioxidant response pathway including enhanced glutathione S transferase activities in DHA-resistant parasites but downregulation of this pathway in the parental strain (60). Of note, culturing of DHA-resistant lines in the absence of DHA led to a revertant phenotype of decreased DHA survival, decreased levels of antioxidant response genes, and a concomitant decrease in glutathione S transferase activities; these were quickly upregulated when revertant lines were re-exposed to DHA (60). PfEXP1 (PF3D7_1121600), a membrane-bound glutathione S-transferase found on the parasitophorous vacuole membrane, was upregulated two-fold in the DHA-tolerant 3b1 parasite strain compared to the parental Dd2 strain, and was further upregulated 2-fold in 3b1 upon DHA exposure (61). In cell-free systems, PfEXP1 can conjugate reduced GSH to hematin, accelerating degradation of the toxic hemoglobin digestion by-product (61). Artesunate inhibits PfEXP1 glutathione transferase activity, likely by alkylation, as endoperoxide-containing antimalarials alkylate PfEXP1 (62), which could functionally inactivate this protein. Pfexp1 was also among the antioxidant response genes upregulated in 3D7 parasites pressured with artemisinin in a separate study (63).

Finally, metabolomics comparisons of trophozoite-stage parasites revealed that artemisinin-resistant parasites Cam3.II K13^R539T and Cam3.II K13^C580Y had two-fold increased levels of glutathione and its precursor gamma-glutamylcysteine compared to the isogenic artemisinin-sensitive counterpart Cam3.II K13^WT (45). These data collectively point to the involvement of the antioxidant response in contributing to artemisinin survival throughout the intraerythrocytic developmental cycle.

Unfolded Protein Response and Ubiquitin Proteasome System

Transcriptomic analyses of more than 1000 clinical isolates from the Tracking Resistance to Artemisinin Collaboration (TRAC) showed that resistant parasites upregulated protein folding chaperones, DNA damage responsive genes, and components of the proteasome (64). Collectively, the data point to activation of the unfolded protein response (UPR). The main goal of UPR activation is to divert cellular resources to manage existing damaged proteins by halting nascent protein production, upregulating chaperones, and increasing proteasomal degradation capacity (65).

Only components to one of the three canonical branches of the eukaryotic UPR transmembrane sensors have been identified in Plasmodium (66). Protein kinase 4 (PK4, PF3D7_0628200), an ortholog of protein kinase R (PKR)-like ER Kinase (PERK), phosphorylates elongation initiation factor 2α (eIF2α) (44). DHA induces P. falciparum eIF2α phosphorylation (10, 44), which is associated with quiescence and decreased sensitivity to DHA and AS (44). Conversely, PK4 inhibition blocks latency and increases DHA and AS susceptibility (44). Artemisinin-sensitive Dd2 harboring K13^WT had no detectable levels of phosphorylated eIF2α in untreated cells but a large increase in phosphorylation upon exposure to 700 nM DHA for 15 minutes (44). In contrast, Dd2^C580Y showed phosphorylated eIF2α in early rings and no further increase upon DHA treatment (44), indicating that artemisinin-resistant parasites have elevated levels of eIF2α phosphorylation and could be engaged in an ongoing stressed program. Whether constant engagement of the UPR is a key tenet beneficial for artemisinin survival requires further investigation.

DHA treatment causes an accumulation of misfolded proteins (10), which could be due to DHA-mediated protein alkylation. The UPR is intimately linked to the ubiquitin proteasome system (UPS), and it is hypothesized that these misfolded proteins must be disposed of by the proteasome for continued parasite survival, as proteasome inhibition is lethal (67, 68), although the proteasome is also involved in other essential processes. DHA also leads to an accumulation of ubiquitinated polypeptides in parasites regardless of artemisinin sensitivity (43). To assess the contribution of ubiquitinated polypeptides to DHA-mediated parasite death, an inhibitor of eukaryotic E1 ubiquitin activating enzymes (compound 1 (5′-O-sulphamoyl-N(6)-[(1S)−2,3-dihydro-1Hinden-1-yl]-adenosine)) was used to inhibit activation of ubiquitin moieties (10). Ring- and trophozoite-stage parasite exposure of up to 20 μM compound 1 alone did not affect parasite viability, although 10 μM was sufficient to deplete ubiquitin levels by western blot visualization (10). Pre-treatment of ring-stage parasites with 20 μM compound 1 for 3 hours followed by exposure to DHA for 3 hours led to decreased DHA susceptibility in K13^WT parasites (10). These data point to the involvement of ubiquitin in DHA-mediated parasite killing.

Proteins destined for destruction are tagged with ubiquitin and brought to the 26S proteasome via ubiquitin shuttle proteins (69). The structure of P. falciparum 20S core particle (CP), which contains the proteolytic subunits β₁, β₂, and β₅ has been solved by cryo-EM (70). The CP in other eukaryotes can be activated by either a 19S regulatory particle (RP) or PA28 (71). P. falciparum CP has been shown to be associated with RP by native gel analysis (72), and PA28 by cryo-EM studies (73). The RP is responsible for substrate recognition, deubiquitination, unfolding, and CP gate opening (71). All subunits of the CP and RP, PA28β, and proteasome interacting partners including p97, ubiquitin shuttle proteins, and proteasome-associated deubiquitinases were identified as an interacting complex by proteomics and biochemistry (72). In mammals, PA28 is involved in degradation of oxidized proteins in a ubiquitin- and ATP-independent manner (71). PfPA28 (PF3D7_0907700) enhances Pf20S-mediated ubiquitin-independent degradation of an intrinsically unstructured protein (73), and could be involved in degradation of oxidized proteins.

Proteasome inhibitors developed for anti-cancer therapy potently inhibit the establishment of liver stages (67), prevent intraerythrocytic development (67), prevent transmission by killing stage V gametocytes (68), and prevent oocyst production within the mosquito (68). These inhibitors also synergize with DHA in vitro and in vivo (43). Recently, there has been an increased effort to identify and develop P. falciparum-specific proteasome inhibitors including vinyl sulfones (70, 74), asparagine ethylenediamines (AsnEDA) (75, 76) epoxyketones (77), and peptide boronates (78). Of note, the vinyl sulfones and a subset of AsnEDAs have been tested and shown to be effective against artemisinin-resistant parasites (70, 74, 75). The epoxyketones synergize with artemisinin in the Dd2 strain. (77). The vinyl sulfones and peptide boronates synergize with DHA against artemisinin-resistant K13^R539T and K13^C580Y parasites (70, 78, 79). Thus, proteasome inhibitors may be an ideal partner drug for artemisinin and related structures to combat multidrug resistance given the prevalence of K13^C580Y in Southeast Asia.

Beyond K13: contribution of other genes to artemisinin resistance in Southeast Asian parasites

Two lines of evidence underscore the contribution of parasite genetic background and non-K13 genes to the artemisinin resistance phenotype: 1) genetic introduction of K13^C580Y into parasites of different geographic backgrounds resulted in varying levels of RSA survival rates (36), and 2) in vitro-generated artemisinin-resistant lines do not always yield K13^PDmut (80, 81, 82). These results highlight the fact that in conjunction with K13, other proteins likely contribute to artemisinin survival.

Cambodian isolates typically carry multiple mutations in proteins associated with drug resistance, including point mutations in pfcrt (chloroquine resistance transporter, PF3D7_0709000), pfdhps (hydroxymethyldihydropterin pyrophosphokinase-dihydropteroate synthase, PF3D7_0810800) and pfdhfr (dihydrofolate reductase, PF3D7_0417200 (83). On this background, Cheeseman et al identified a 35-kb region on chromosome 13 that was associated with delayed parasite clearance times (83). Within this region lie the ER chaperone GRP170 (PF3D7_1344200), lipoate synthase (PF3D7_1344600), and thioredoxin 2 (PF3D7_1345100) (83), which function in cell stress responses, lipoic acid salvage/biosynthesis pathway, and oxidative responses, respectively. SNPs in a conserved Plasmodium protein of unknown function (PF3D7_1344700) and zinc finger protein (PF3D7_1344300) showed the strongest association with delayed clearance times in the region of selection (83).

A separate study that genotyped fast-, medium-, and slow-clearing parasites from Bangladesh, Thailand, and Cambodia using a P. falciparum SNP Affymetrix array found that a region adjacent to the 35-kb region mentioned above was found to be associated with slow-clearing parasites. Specifically, mutations in the coding and 3’ untranslated region of Rad5 (Pf3D7_1343400), the 3’ untranslated region of DNA polymerase δ-catalytic subunit (PF3D7_1017000), and a guanine nucleotide exchange factor (PF3D7_1417400) were strongly associated with slow-clearing parasites (84). Subsequently, genome-wide analyses of parasites from these three countries revealed that slow-clearing Cambodian parasites strongly associated with K13 (85), confirming Ariey et al.’s report. In addition, genes with SNPs in linkage disequilibrium windows associated with delayed parasite clearance times included pfcrt, a putative retrieval receptor for ER membrane proteins (PF3D7_0903100), and K13 (85).

A GWAS study examining parasite isolates obtained from patients in Bangladesh, Myanmar, Thailand, Vietnam, Laos, Cambodia, Democratic Republic of Congo, and Nigeria recruited into TRAC and a study by the National Institutes of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH) showed that although the most significant association with artemisinin resistance is SNPs in K13, specifically C580Y, other strong associations included SNPs in pfarps 10 (apicoplast ribosomal protein S10, PF3D7_1460900), pfmdr2 (multidrug resistance protein 2, PF3D7_1447900), pffd (ferredoxin, PF3D7_1318100), pph (putative protein phosphatase, PF3D7_1012700), pfpib7 (putative phosphoinositide-binding protein, PF3D7_0720700), pfcrt, and two genes of unknown function (PF3D7_1322700, PF3D7_1451200) (86). SNPs in pffd, pfcrt, pfarps10 and pfmdr2 were shown to be associated with resistant founder populations and a K13^PDmut (86), implying it is upon this genetic background that K13 mutations are likely to emerge.

A longitudinal genomic survey of parasite isolates from Northwest Thailand conducted from 2001–2014, a period of time during which artemisinin resistance emerged, revealed genes that co-evolved polymorphisms in conjunction with the evolution of K13^C580Y. Among these are proteins involved in the phosphatidyl signaling pathway and vesicular trafficking (phosphatidylinositol 4-kinase (PI4K, PF3D7_0419900) and Sec14-domain containing protein (PF3D7_0626400)), ubiquitin conjugation and deconjugation enzymes (Hrd3 (PF3D7_1448400) and UBP1 (Pf3D7_0104300)), the chaperone Hsp40 (PF3D7_1437900), a sentrin-specific protease 2 (SENP2, PF3D7_0801700), and kelch10 (PF3D7_1022600) (87).

Parasites from the China-Myanmar border displaying increased DHA and AM IC₅₀ values showed a strong correlation with a ~905 kb segment on chromosome 10 (88). This region encompasses DNA polymerase δ-catalytic subunit (88), of which SNPs in this gene’s 3’ UTR was previously identified by (84) to associate with delayed parasite clearance times. The neighboring genes autophagy-related protein 18 (ATG18, PF3D7_1012900) and NLI interacting factor-like phosphatase (NIF4, PF3D7_1012700) were also in the region identified (88). Analysis of the same parasites using RSA values instead of IC₅₀ values identified Rad5, elongation factor G (PF3D7_0602400), and geranylgeranyltransferase (PF3D7_0602500) to be signficantly associated with artemisinin resistance (88).

Although most artemisinin-resistant parasites from Cambodia included in the TRAC study harbored K13^C580Y, two of the 36 culture-adapted parasites with no K13 mutations demonstrated an in vivo parasite clearance half-life of less than 5 hours but an in vitro RSA value of greater than 1% (89). The increased RSA values could be attributed to mutations in genes such as pfdhfr, pfdhps, and pfmdr2, which are known to modulate resistance to other antimalarials (89).

Point mutations and copy number variation in the DV-resident transporter PfMDR1 mediate parasite susceptibility to multiple quinolines and arylaminoalcohols (90, 91, 92) There are five prevalent mutations in PfMDR1: three at the carboxy-terminus (S1034C, N1042D, and D1246Y) and two at the amino-terminus (N86Y and Y184D) (90). The S1034C + N1042D + D1246Y triple point mutation prevalent in South America was inserted into P. falciparum strains 3BA6 and GC03, resulting in a ~2–4-fold decrease in artemisinin IC₅₀ values (93). Examination of drug sensitivity of isogenic NF10 and KC5 strain parasites genetically edited at position 86 or 184 revealed that PfMDR1^N86 parasites display a modest but significant ~1.5 fold increase in DHA IC₅₀ compared to those harboring PfMDR1^N86Y (94). Of note, PfMDR1^N86Y was absent from the clinical isolates analyzed from Southeast Asia (94). In addition, increased pfmdr1 copy number was associated with increased artemisinin resistance in vitro (91) and in vivo (95). A separate study showed that in vitro DHA resistance selections yielded resistant parasites with increased pfmdr1 copy number and expression, which was subsequently lost upon removal of DHA exposure (60). Loss of pfmdr1 amplification, in addition to decreased levels of genes in the antioxidant response pathways, correlated with increased susceptibility to DHA (60). Collectively, the data show that pfmdr1 expression levels and haplotypes modulate artemisinin sensitivity but are likely not key determinants of artemisinin resistance. Note that parasites harboring K13^PDmut have a genetic background that includes mutations or copy number variation in pfmdr1 (94).

The in vitro-generated F32-ART5 evolved mutations in an RNA polymerase II subunit (PF3D7_0110400) at the same time as K13^M476I, and upon prolonged incubation with artemisinin subsequent mutations were detected in falcipain 2a (FP2a, PF3D7_1115700), protein kinase 7 (PK7, PF3D7_0213400), gamete antigen 27/25 (PF3D7_1302100), and two proteins of unknown function (PF3D7_1459600, PF3D7_1464500) (35). Although these additional mutations did not increase RSA survival rates (35), they may represent compensatory mutations that contribute to parasite fitness.

In vitro genetically modified 3D7 strain ΔFP2a parasites, which have reduced hemoglobin digestion in the early and mid-trophozoite stages (96), demonstrated increased resistance to a 3-hour pulse of artemisinin (5, 17) but not DHA (17) at the trophozoite stage. Exposure to the cysteine protease inhibitor E64d restored viability to artemisinin- but not DHA-treated 3D7 trophozoite stage parasites (17), suggesting there may be differences in parasite activation of these drugs. Delineation of artemisinin resistance in early and late ring stages of ΔFP2a was less straightforward. ΔFP2a parasites were more resistant to artemisinin at early ring stages (1.5 hpi) but were sensitized at late ring stages (6 hpi) compared to the parental 3D7 strain (17). Given the necessity of hemoglobin digestion in parasite survival, it is perhaps unsurprising that although ΔFP2a trophozoite stage parasites display phenotypic digestive vacuole defects and decreased hemoglobin digestion, they do not display defects in hemoglobin digestion at later stages, and do not display a growth defect (96). FP3, which is expressed during late trophozoite and schizont stages, likely plays a compensatory role in these parasites, as fp3 transcripts increased in ΔFP2a parasites (96).

Nevertheless, profiling of 140 parasite isolates collected from the China-Myanmar border from 2004–2011 revealed 25 distinct haplotypes with multiple mutations in FP2a, which can be categorized into 8 haplotype groups (97). All parasites regardless of FP2a haplotype group displayed significantly decreased levels of cysteine protease activity as assayed by z-LR-AMC hydrolysis compared to FP2a WT parasites. RSA values for FP2a mutants ranged from 1–6 % survival rates (compared to 0.5% of FP2a WT parasites), but these values did not correlate with FP2 activity (97), indicating there may be other compensatory mechanisms contributing to fitness and survival in RSAs.

K13 and artemisinin resistance in African and South American parasites

Non-synonymous SNPs in the K13 propeller domain (K13^PD) occurs at a less than 10% frequency in Africa, South America, and Oceania (40). For example, the K13^C580Y mutation found prevalently in Southeast Asia (40, 41) was only found in 5% of parasites obtained from Guyana in 2010 (98). Genomic surveillance from 2010–2014 revealed that < 2% of parasites profiled from the Democratic Republic of Congo, Gabon, Ghana, Kenya, Mali, and Uganda had propeller domain mutant K13^A578S, which is the most frequently reported allele in Africa (99). No clinical outcomes or resistance phenotypes are available for the isolates described in these studies, and these areas continue to respond to artemisinin treatment (1, 98, 99). However, it has been reported that parasites from Uganda with K13^A578S demonstrate increased parasite clearance times (80 hours compared to 45 hours for K13^WT infections) (29).

Blood samples from returning Chinese migrant workers who had been in Ghana revealed several mutations in K13^PDmut including K13^C580Y (100). It is unknown whether these workers received treatment, and RSAs with these isolates were not performed; thus, it is unclear if these infections are artemisinin resistant. Parasites isolated from Chinese migrant workers returning from 27 countries in Africa prior to treatment revealed K13^PDmut-harboring parasites, but the resistance status of these parasites is unknown (101). There has been one report of an artemisinin-resistant K13^M579I parasite from a Chinese migrant worker in Equatorial Guinea (28). Parasites were still detectable after 3 days of DHA-PQP, and whole genome sequencing followed by principal component analysis demonstrated its African origins (28).

Yet, K13^PDmut do not always seem to correlate with artemisinin resistance in Africa. While there were three distinct K13^PDmut found in parasite isolates from Senegal in 2015–2016 (1.7% of parasites examined), all three parasite infections were cleared within three days (30). On the other hand, 9 of 119 patients remained parasitic on day 3, but all harbored K13^WT parasites (30). Interestingly, examination of parasites obtained from patients in Tanzania showed that parasites whose k13 transcripts decreased after AM-LF treatment significantly correlated with increased parasite clearance times (102). None of these parasites had mutations in the propeller domains (102). Thus, decreasing k13 levels may be a mechanism for African parasites to survive artemisinin. Of concern, studies in Suriname, which borders French Guiana, Brazil, and Guyana, show that 31% of patients still had detectable parasites in peripheral blood smears at day 3 after treatment with AM-LF when assessed in 2011 compared to only 2% when assessed five years prior (32). Genetic analyses on these parasites were not performed and information about K13 status is not available (32). These cases demonstrate that mutations in the K13 propeller domain alone are not sufficient to confer artemisinin resistance in the absence of other genes necessary for a resistance phenotype. With this perspective, studies examining in vitro generated artemisinin resistance and the contribution of candidate genes could yield significant insight into resistance-conferring haplotypes in non-Southeast Asian regions.

Non-K13-mediated artemisinin resistance in African and South American parasites

Transcriptomics analyses of Gambian FCR3 parasites exposed to a sub-lethal pulse of AS revealed altered transcript levels in approximately 400 genes (103). Of these, a tryptophan-rich protein of unknown function (PArt, PF3D7_1002200) was identified to have the greatest increase in expression, and AS treatment led to increased protein expression (104). Though overexpression of PArt did not alter AS IC₅₀ as assessed by a [³H]-hypoxanthine uptake inhibition assay, the authors noted PArt-overexpressing parasites exhibited a less severe AS-induced developmental delay when ring stage parasites were pulsed for 3 hours with a sub-lethal dose of AS (104), which may indicate a supporting role in artemisinin resistance.

In vitro artemisinin selection (105) and subsequent linkage analyses of the rodent malaria P. chabaudi identified a I568T mutation in the μ subunit of the AP2 endocytic adaptor complex (PcAP2μ, PCHAS_143590) (80). Modeling studies show that I568T moderates but does not abrogate binding of PcAP2μ to the recognition motif in cargo protein (80). Introduction of I592T into PfAP2μ, the analogous mutation of PcAP2μ^I568T, led to ~10% survival in RSAs compared to the parental 3D7 strain survival of 1% (48). In Kenya, São Tomé, and Rwanda, the nearby mutations PfAP2μ^S160N and PfAP2μ^S160T were associated with increased parasite survival following ACT treatment failure (80, 106). Surprisingly, 3D7 parasites genetically engineered to express PfAP2μ^S160N did not demonstrate artemisinin resistance via RSA (48), indicating that this mutation may require particular African parasite genetics to confer artemisinin resistance. Knock-sideways of PfAP2μ significantly reduced endocytic uptake, as indicated by a decrease in DV size (55). Together, these data suggest that artemisinin resistance-associated mutations in PfAP2μ could decrease hemoglobin uptake, limiting the amount of free heme available to activate artemisinin.

Analyses of P. chabaudi exposed to pyrimethamine followed by either chloroquine or artemisinin revealed that surviving parasites harbored SNPs V2728F or V2697F in the deubiquitinating enzyme ubiquitin specific protease 1 (PcUBP1, PCHAS_0207200), respectively (81). Subsequent analyses determined that parasites harboring PcUBP1^V2728F or PcUBP1^V2697F are resistant to artemisinin (107). 3D7 strain parasites genetically engineered to encode PfUBP1^V3275F, the analogous P. falciparum mutation to PcUBP1^V2728F, displayed artemisinin resistance with an 8.5% survival rate in RSAs (48). However, PfUBP1^V3306F, the analogous mutation to PcUBP1^V2697F, showed similar RSA values to the non-edited 3D7 strain parasites (48). PfUBP1 interacts with K13 as shown by DiQ-BioID, and knock-sideways of PfUBP1 decreases endocytic uptake (55). These in vitro studies of UBP1 mutant-mediated artemisinin resistance are strengthened by observations that parasites obtained following ACT treatment failure in Kenya harbored a E1528D mutation in PfUBP1 (106), and that ex vivo studies using P. falciparum isolates from Kenya demonstrated that PfUBP1^K837R increased parasite resistance to DHA (108). Note that these PfUBP1 mutations were not found in artemisinin-resistant strains from Western Cambodia (109), which could indicate that mutations in distinct proteins emerge on disparate parasite genetic backgrounds to mediate artemisinin resistance.

In vitro drug selection with recent isolates from Pikine and Thiès, Senegal, produced DHA-resistant parasites (82). Whole genome sequencing identified polymorphisms in seven genes; of these, two genes were mutated in both resistant strains: coronin (PF3D7_1251200), and a conserved Plasmodium protein of unknown function (Pf3D7_1433800) (82). PfCoronin possesses a seven-bladed propeller domain composed of WD-40 repeats and β-propeller folds in its N-terminus, reminiscent of the β-propeller folds of Kelch domains in K13 (82). Introduction by CRISPR-Cas9-mediated gene editing of either PfCoronin G50E or the double mutation R100K + E107V, all of which lie in the WD-40 domain, resulted in a 5–9% survival rate in RSAs (82). PfCoronin is expressed late in erythrocytic development, localizes to the pellicle region of invasive merozoites during erythrocyte entry, and is involved in F-actin organization (110). It is likely important for erythrocyte invasion, but is dispensable to parasite growth (111). Note that these particular artemisinin resistance-conferring PfCoronin mutations occurred in parasites that harbored PfCoronin^S183G and K13^K189T, which do not by themselves confer resistance as measured by RSA (82). Importantly, these West African parasites do not have the polymorphisms in pfarps10, pfmdr2, pffd, or pfcrt that Miotto et al. found in Cambodian parasites associated with artemisinin resistance (82, 86). These data highlight the importance of parasite background genetics for the evolution of artemisinin resistance-conferring polymorphisms, and that findings true for Cambodian parasites may not translate to African parasites.

Other than in vitro selection studies, targeted approaches examining the contribution of particular parasite proteins to artemisinin resistance have also been investigated. Thapsigargin, a well-characterized inhibitor of human SERCA (sarco/endoplasmic reticulum Ca²⁺-dependent ATPase) (112), and artemisinin are both sesquiterpene lactones. Artemisinin inhibits activity of the P. falciparum SERCA ortholog PfATP6 (PF3D7_0106300) expressed heterologously in Xenopus oocytes (113). However, further studies showed that parasites expressing PfATP6^L263E, a mutation near the thapsigargin binding pocket, did not alter sensitivity to artemisinin, DHA, or AS in standard 72 hour growth inhibition assays (114). It would be interesting to see whether this mutation affects survival in the newer RSAs. In French Guiana and Senegal, polymorphisms in PfATP6 were associated with reduced susceptibility to artemisinin derivatives (115), though no resistance-associated polymorphisms were found in Cambodia (109). PfATP6 alkylation was detected in two artemisinin-interactome studies (8, 72), but not a third study (9), suggesting that PfATP6 could be a nonspecific target of artemisinin. Overall, the data suggest that PfATP6 may interact with artemisinin, although how mutations in this gene modulates artemisinin resistance is yet to be determined.

In summation, polymorphisms in four genes have been confirmed by gene editing to confer artemisinin resistance: K13 (36, 42), PfCoronin (82), PfAP2μ (48), and PfUBP1 (48). Polymorphisms in K13^PDmut is the most established cause for artemisinin resistance, as this has also been confirmed in clinical settings. For a comprehensive review and analysis of K13 gene polymorphisms and artemisinin resistance in malaria-affected countries the reader is referred to (116). Mutations in K13^PDmut emerged in Cambodian parasites with genetic backgrounds that harbor mutations in known antimalarial resistance modulating genes, and genes involved in hemoglobin digestion, antioxidant response, DNA repair, and cell stress pathways. Genes involved in these pathways, independently of K13^PDmut, have also been shown to confer artemisinin resistance in vitro (Figure 3). Further examination of parasites of different geographic locations could reveal additional genes involved in artemisinin resistance, and the evolution of these genetic factors should be monitored.

Figure 3.

Mechanisms of artemisinin activation and resistance. Hemoglobin is imported from the host red blood cell and transported to the parasite digestive vacuole where it is proteolytically cleaved, releasing free heme. Heme is toxic to the parasite and is thus converted in a series of steps to the inert hemozoin crystal (11), indicated as yellow rectangles. Trace amounts of heme within the parasite can activate artemisinin (14). Bioactivation of artemisinin leads to non-selective widespread alkylation of parasite proteins (7, 8, 9) and heme (19, 20), which is thought to interfere with essential parasite processes. Artesunate (3) and heme-artemisinin adducts (4) inhibit hemozoin polymerization, which could contribute to heme-induced parasite toxicity. In addition, DHA has been shown to lead to increase in ROS (5, 6). DHA leads to inhibition of proteasomes (10), indicated as a set of light and dark green ovals, which are presumed to be required to dispose of toxic aggregates of alkylated or ROS-damaged proteins. P. falciparum possesses a complete 26S proteasome system (72) in addition to PA28 activator (73). Parasite artemisinin resistance can be mediated by a decrease in artemisinin activation by: (a) decreased hemoglobin import (48, 55), or (b) decreased hemoglobin digestion (17, 45, 47, 97). Alternatively, parasites can counter artemisinin-mediated toxicity by increasing: (c) antioxidants (45, 60, 63), indicated by a purple triangle, or (d) activation of the unfolded protein response (UPR) (10, 44, 64), characterized by eIF2α phosphorylation (10, 44), leading to global translational attenuation (10) with selective upregulation of chaperones and proteasome subunits (64). The UPR is triggered when misfolded proteins accumulate and titrate BiP away from PK4, leading to dimerization and autotransphosphorylation and subsequent eIF2α phosphorylation (44). ARE, antioxidant response element; ART, artemisinin and its derivatives; BiP, binding immunoglobulin protein; eIF2α, elongation initiation factor 2α; Fe²⁺-FPIX, ferroprotoporphyrin; Fe³⁺-FPIX, ferriprotoporphyrin; PK4, protein kinase 4; ROS, reactive oxygen species.

Beyond artemisinin

Despite the potency of artemisinin, efforts to improve on its limitations led to the development of synthetic ozonide antimalarials (117, 118) (Figure 4A) that can circumvent limitations of supply-dependent Artemisia annua harvest and extraction costs, short in vivo 1–2 hour half-lives of artemisinins, and artemisinin resistance. For a comprehensive review of ozonide antimalarial activity including detailed pharmacokinetic/pharmacodynamic profiles the reader is directed to (119). The first-generation trioxolane OZ277 (arterolane) (117) demonstrates improved stability and an increased in vivo half-life of 2–3 hours (120, 121). OZ277 is licensed in India and seven African countries as Synriam™, a three-dose combination therapy with PQP. In contrast to OZ277, OZ439 (artefenomel) resists premature iron (II) and heme-mediated degradation (118). Phase II clinical trials demonstrated that OZ439 is potent against both P. falciparum and P. vivax malaria with a long in vivo half-life of 46–62 hours (122). This promising finding begged the question of which antimalarial to pair OZ439 with for therapeutic use. Several partner drugs were tested in clinical trials, including the parasite pyrimidine biosynthesis inhibitor DSM265 (123), and the quinolines PQP and ferroquine (FQ, also known as SSR97193) (124) (Figure 4B), which inhibit parasite-mediated heme detoxification (125, 126).

Figure 4.

(A) Synthetic ozonide antimalarials. The cis-8′-alkyl group, indicated in green, and cis-8′-phenyl substituent, indicated in blue, on OZ277 and OZ439, respectively, protect the endoperoxide bridge, indicated in red, from premature cleavage. (B) Ferroquine is an organometallic drug, consisting of a 4-aminoquinoline group, a ferrocene group, and an alkamine group. This compound has completed Phase II clinical trials with OZ439.

A Phase Ib clinical trial evaluating a single dose regimen with OZ439 and the bis-4-aminoquinoline PQP has recently been completed (ClinicalTrials.gov Identifier: NCT03542149). Unfortunately, PQP resistance has emerged in regions of Southeast Asia where DHA-PQP was the standard care of treatment (127, 128). Amplification of plasmepsin II and III genes (PF3D7_1408000, PF3D7_1408100), and a E415G mutation in an exonuclease (PF3D7_1362500) has been associated with PQP resistance in Cambodian clinical isolates (129, 130, 131, 132). In addition, PQP resistance can be conferred by novel mutations in pfcrt (133). These recent findings could curtail the use of OZ439-PQP therapy in the future.

Phase IIa/b clinical trials testing OZ439 and FQ as a single dose regimen has recently been completed (ClinicalTrials.gov Identifiers: NCT03660839, NCT02497612). FQ, a ferrocene-4-aminoquinoline hybrid, is highly potent (IC₅₀ below 30 nM) against >500 clinical isolates from Southeast Asia and Africa regardless of chloroquine sensitivity (134) and against 80 clinical isolates from Cambodia harboring either K13^WT or K13^C580Y (135). Parasites with amplification of plasmepsin II or amplification of pfmdr1 in addition to harboring K13^C580Y demonstrated FQ IC₅₀ of 17 nM and 9 nM, respectively, compared to a FQ IC₅₀ of 13 nM for K13^C580Y parasites without plasmepsin II and pfmdr1 amplifications (135). These results demonstrate that FQ remains effective regardless of underlying genes that significantly contribute to artemisinin, piperaquine, and mefloquine resistance in the Southeast Asian region. Moreover, in vitro evolution studies demonstrated that parasites were unable to acquire resistance to FQ when starting with 3 × 10⁸ parasites of W2 strain (136) or up to 2 × 10⁹ Dd2 strain parasites (Ng, C.L. and Fidock, D.A., unpublished data). These favorable characteristics lend optimism for the success of OZ439-FQ combination therapies. One shortcoming is that FQ inhibits OZ439 solvation in milk (137), which is necessary for improved OZ439 bioavailability (138). However, an OZ439 nanoparticle formulation (139) could overcome these hurdles by obviating the need for administering OZ439 with milk. This formulation is stable up to 6 months at 40°C and 75% relative humidity (140), which mimics the conditions in South America, India, Southeast Asia, and Africa. OZ439 nanoparticles could provide an inexpensive means for feasible drug administration across hot, humid climates where disease is endemic.

Although there are differences in parasite behavior in response to artemisinins and the synthetic ozonides, there are overall similarities. OZ439 is activated primarily by heme, generating secondary carbon centered free radicals (141) which alkylate parasite proteins (62, 142) and heme (143). Indeed, using ozonide click chemistry, the alkylation profile of OZ439 was shown to share 85% similarity to that of artemisinin (62). Importantly, with a half-life of 46–62 hours, OZ439 allows effective drug exposure to all asexual parasite stages (122). This decreases the probability of sub-lethal drug exposure and reduces the likelihood of parasite drug resistance, lending traction to the idea of replacing artemisinins with OZ439. In vitro, cross resistance has not been observed with parasites harboring K13^R539T (144, 145) or the prevalent K13^C580Y mutation (145). When tested against parasites harboring the rare K13^I543T mutation, a modest amount of cross-resistance was observed, with RSA values of 3.3% compared to 0.7% for K13^WT (145). Note that since OZ439 has a significantly longer half-life than artemisinin, traditional RSAs in which parasites are only exposed to a 3–6 hour drug treatment may not accurately measure the in vivo resistance profile of OZ439. Experiments utilizing extended drug pulses that mimic in vivo pharmacokinetics show that in contrast to DHA and OZ277, exposure of 34 nM OZ439 or 19 nM to OZ609 (a newer compound in this class) for 48 hours similarly led to < 1% survival rates of both the artemisinin-resistant clinical isolate Cam3.I K13^R539T and its isogenic sensitive strain Cam3.I K13^WT (49). Nevertheless, the possibility of drug resistance should remain a consideration and partner drugs must be strategically selected. An ideal partner drug would synergize with OZ439, select for mutations that antagonize resistance, and be difficult to develop resistance against on its own. P. falciparum-specific proteasome inhibitors demonstrated a low propensity of resistance and demonstrated potent synergy with OZ439 (79). Combination therapy of P. falciparum-specific proteasome inhibitors and OZ439 in in vivo experiments and clinical trials in the future is eagerly awaited.

For ACS Infectious Diseases Wikipedia:

Delayed parasite clearance in response to artemisinin, the cornerstone of first-line artemisinin-based combination therapies, has been detected in Asia, Africa, and South America. Although resistance appears to be mediated by different genetic loci in parasites of distinct regions, resistance mechanisms overlap and are associated with alterations in hemoglobin uptake, hemoglobin digestion, antioxidant response, DNA repair, and cell stress pathways.

Abbreviations

ACT

artemisinin-based combination therapy

ADQ

amodiaquine

AM

artemether

AS

artesunate

Cas9

CRISPR-associated protein 9

CP

core particle

CRISPR

clustered regularly interspaced short palindromic repeat

DHA

dihydroartemisinin

eIF2α

elongation initiation factor 2α

Fe²⁺

ferrous iron

Fe³⁺

ferric iron

FQ

ferroquine

hpi

hours post-invasion

IC₅₀

half-maximal inhibitory concentration

K13^PD

K13 propeller domain

K13^PDmut

K13 propeller domain mutant

K13^WT

wild type K13

LF

lumefantrine

MF

mefloquine

PI3K

phosphoinositide 3-kinase

PI3P

phosphatidylinositol 3-phosphate

PK4

protein kinase 4

PQP

piperaquine

PYR

pyronaridine

RBC

red blood cell

ROS

reactive oxygen species

RP

regulatory particle

RSA

ring-stage survival assay

SNP

single nucleotide polymorphism

SP

sulfadioxine-pyrimethamine

TRAC

Tracking Resistance to Artemisinin Collaboration

UPR

unfolded protein response

ZFN

zinc finger nuclease