A Proteasome Mutation Sensitizes P. falciparum Cam3.II K13^C580Y Parasites to DHA and OZ439

Abstract

Artemisinin-based combination therapies (ACTs), the World Health Organization-recommended first-line therapy for uncomplicated falciparum malaria, has led to significant decreases in malaria-associated morbidity and mortality in the past two decades. Decreased therapeutic efficacy of artemisinins, the cornerstone of ACTs, is threatening the gains made against this disease. As such, novel therapeutics with uncompromised mechanisms of action are needed to combat parasite-mediated antimalarial resistance. We have previously reported the antimalarial activity of Plasmodium falciparum-specific proteasome inhibitors in conjunction with a variety of antimalarials in clinical use or in preclinical investigations and of proteasome mutants generated in response to these inhibitors. Here, we discover that despite harboring K13^C580Y, which has conventionally mediated artemisinin resistance in vitro as measured by increased survival in ring-stage survival assays (RSA), the Cam3.II strain parasites of Cambodian origin that have acquired an additional mutation in the proteasome display increased susceptibility to DHA and OZ439. This discovery implicates the proteasome in peroxide susceptibilities and has favorable implications on the use of peroxide and proteasome inhibitor combination therapy for the treatment of artemisinin-resistant malaria.

Graphical Abstract

Dihydroartemisinin (DHA) and OZ439 cause an accumulation of misfolded proteins in P. falciparum. In order to survive these antimalarials, parasites rely on the proteasome to degrade damaged proteins (left, top). Cam3.II K13^C580Y parasites with a mutation in the β2 proteasome subunit, indicated by red circles (left, bottom), display increased sensitivity to DHA and OZ439 as assayed by growth inhibition studies (right). Our data directly implicate the proteasome in peroxide susceptibilities. For simplicity, the graphic is shown with DHA as a representative of peroxides.

Malaria remains a significant global health concern, responsible for a recorded 229 million cases and 409,000 deaths in 2019 ¹. There are five species of the eukaryotic protozoan parasite Plasmodium that cause malaria in humans, of which P. falciparum is the deadliest. Artemisinin-based combination therapies (ACTs), comprised of a fast-acting artemisinin derivative and a long-lasting partner drug, are largely responsible for the substantial (44%) decline in malaria-related deaths over the past two decades ¹. Alarmingly, resistance to artemisinins has developed in Cambodia, Thailand, Laos, Myanmar, and Vietnam, countries that comprise the Greater Mekong Subregion (GMS) ^2–4. Outside of Southeast Asia, several cases of artemisinin resistance have been reported in Equatorial Guinea ⁵, Uganda ⁶, Rwanda ⁷, Senegal ⁸, India ⁹, and the Guiana Shield ^{10, 11}. In vivo, artemisinin resistance is defined as parasite clearance half-life greater than 5 h following treatment with an artemisinin or an ACT ³ and can result in treatment failure if WHO-recommended ACT administration schedules of multiple dosing are not followed. Artemisinin resistance is strongly associated with point mutations in the propeller domain of kelch 13 (K13, PF3D7_1343700) ^{3, 12}, and a handful of K13 mutations including M476I, Y493H, R539T, I543T, and C580Y have been experimentally confirmed to confer resistance ^{13, 14}. The most prevalent clinical mutation in the GMS is K13 C580Y ^{15, 16}. A predominant theory of K13-mediated artemisinin resistance is that parasite hemoglobin import and digestion is reduced in K13 propeller domain mutants (K13^PDmut) ^17–19, thus reducing the amount of free heme available to activate artemisinins.

Upon activation by heme, artemisinins nonspecifically alkylate adjacent proteins ^{20, 21}, causing widespread protein damage. This suggests that parasites that survive artemisinin treatment possess increased capacity to restore proteostasis. In eukaryotes, the ubiquitin proteasome pathway serves as the primary mechanism to degrade and recycle misfolded and damaged proteins ²². Substrates tagged with K48- or K11-linked ubiquitin moieties are targeted for proteasome-mediated degradation ²³. The 26S proteasome has two components: the 19S regulatory particle (RP) and the 20S core particle (CP). The RP is responsible for substrate recognition, deubiquitination, unfolding, and insertion into the CP ²². The CP, composed of two heptameric rings of β subunits sandwiched by heptameric rings of α subunits, carries out the catalytic function ²². β1 has caspase-like activity, cleaving after acidic residues; β2 has trypsin-like activity, cleaving after basic residues; and β5 has chymotrypsin-like activity, cleaving after non-polar residues ²⁴. In P. falciparum, the CP has been visualized by cryoelectron microscopy ^{25, 26} and all components of the CP and RP have been biochemically identified as an interacting complex ²⁷.

Chemical inhibitors developed against human proteasomes have antimalarial activity across the lifecycle of Plasmodium parasites in the liver stages, blood stages, and mosquito stages ^{28, 29}. Specifically, lactacystin prevents intraerythrocytic growth as well as the establishment of liver stages ²⁸. Epoxomicin and ALLN prevent intraerythrocytic development and maturation to stage V gametocytes, and incubation of these inhibitors with stage V gametocytes prior to mosquito feeding inhibited oocyst production within the mosquito midgut ²⁹. Dihydroartemisinin (DHA), the active metabolite of all clinical artemisinins, synergizes with epoxomicin in vitro and with carfilzomib in a mouse model ³⁰.

In the last five years, parasite-specific proteasome inhibitors of various chemotypes have been developed ^{25, 31–35}. We have recently reported antimalarial activity of the covalent vinyl sulfone inhibitors WLW-vs (WLW), which inhibits the P. falciparum β2 subunit, and WLL-vs (WLL), which inhibits both the β2 and β5 subunits of the P. falciparum proteasome ^{25, 36}. These inhibitors demonstrate favorable characteristics of an antimalarial candidate, including a high therapeutic index, high potency against artemisinin-resistant parasites with IC₅₀ values in the low nanomolar range, a low propensity of parasite resistance acquisition, and potent synergy with the clinical drug DHA and OZ439, which is in clinical trials ^{25, 36}. Although it was difficult to generate parasite resistance to these proteasome inhibitors, we nonetheless did manage to isolate parasites with mutations in the 26S proteasome that could survive WLL or WLW ³⁶. 78% of WLW-selected parasites were found to have a mutation in the β2 subunit, while 50% of WLL-selected parasites were found to have a mutation in the β5 subunit ³⁶. In this study, we investigate the peroxide susceptibility of parasites of Cambodian origin (Cam3.II strain) harboring mutations at the K13, β2, and β5 loci.

Results

A Mutation in the β2 Proteasome Subunit Sensitizes Cam3.II K13^C580Y Parasites to DHA

Susceptibility to DHA was measured using the ring stage survival assay (RSA_0–3h, also referred to as RSA) ³⁷ and determination of ring-stage half-maximal inhibitory concentrations (ring IC₅₀) using dose-response inhibition assays, as described in the literature ³⁸. A 1% survival threshold in the RSA is used to designate artemisinin sensitivity vs resistance in vitro and correlates with fast-clearing and slow-clearing parasites, respectively ^{12, 14, 37}. RSA and IC₅₀ values measure related but distinct features of artemisinin susceptibility, and genome-wide association studies (GWAS) using IC₅₀ values or RSA values identified different genes with strong associations to that particular measurement ³⁹, indicating that there is a qualitative difference between these two measurements. To examine the susceptibility of proteasome mutants to DHA, RSAs were conducted on tightly synchronized 0–3_hpi ring stage parasites isolated from the following parasites: Cam3.II K13^WT, Cam3.II K13^C580Y, Cam3.II K13^C580Y β2 C31Y, Cam3.II K13^C580Y β2 C31F, and Cam3.II K13^C580Y β5 A20S (Figures 1, S1, and S2). Cam3.II K13^WT and Cam3.II K13^R539T strains were included as internal experimental controls for artemisinin susceptibility and resistance (Figure S1A–C) as they display known low and high RSA values, respectively. Cam3.II K13^WT, Cam3.II K13^R539T, and Cam3.II K13^C580Y parasites displayed survival rates of 0.7%, 34.3%, and 11.4%, respectively (Figures S1 A, B), which is consistent with previously reported values ^{14, 38}. Cam3.II K13^C580Y β2 C31F and Cam3.II K13^C580Y β5 A20S parasites displayed survival rates similar to the parental Cam3.II K13^C580Y of 11.4% and 15.7%, respectively (Figure 1A). In contrast, Cam3.II K13^C580Y β2 C31Y parasites displayed a 2-fold decrease in survival rate (6.1%) relative to its parent (Figure 1A), although this was not statistically significant (p = 0.056). In dose-response inhibition assays, the Cam3.II K13^WT, Cam3.II K13^R539T, and Cam3.II K13^C580Y parasites had similar ring-stage IC₅₀ values (Figure S1C), as previously reported ³⁸. The β5 A20S mutant displayed similar IC₅₀ values to its parent (15.2 nM vs. 19.5 nM) (Figure 1B). In contrast, both β2 C31Y and β2 C31F mutants were sensitized to DHA and had IC₅₀ values of 7.3 nM (p = 0.016) and 10.4 nM (p = 0.19), respectively (Figure 1B). Note that DHA IC₅₀ values in the β2 mutants, which harbor K13^C580Y, are even lower than the 18.1 nM IC₅₀ value for the artemisinin-sensitive Cam3.II K13^WT parasite (Figure 1B). Dose-response curves are shown in Figure S2A.

Figure 1. Cam3.II K13^C580Y β2 proteasome mutants display increased sensitivity to DHA.

(A and B) 0–3_hpi ring stages and (C) 26–30_hpi trophozoite stages isolated from Cam3.II K13^WT, Cam3.II K13^C580Y, Cam3.II K13^C580Y β2 C31Y, Cam3.II K13^C580Y β2 C31F, and Cam3.II K13^C580Y β5 A20S parasites were exposed to a range of DHA concentrations for 3 h, the drug was washed out, and then parasitemia was assessed approximately 66 h later. (D) Asynchronous cultures were exposed to a range of DHA for 72 h. Bar graphs depict (A) quantification of percent survival at 700 nM DHA, and (B-D) mean IC₅₀ values ± SEM. Values were calculated from at least three biological replicates. Statistical significance was examined between proteasome mutants and Cam3.II K13^WT or Cam3.II K13^C580Y parasites using a Mann-Whitney U test. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; n.s. = not significant.

Since the proteasome is upregulated during the trophozoite stages ^{40, 41}, we repeated the experiments outlined above but used tightly synchronized 26–30_hpi trophozoites (Figure 1C). Cam3.II K13^WT, Cam3.II K13^R539T, and Cam3.II K13^C580Y parasites demonstrated similar IC₅₀ values (Figures S1D, E), as expected. Cam3.II K13^C580Y and Cam3.II K13^C580Y β5 A20S parasites had similar DHA trophozoite IC₅₀ values of 12.4 nM and 13.7 nM, respectively (Figure 1C). In contrast, the β2 C31Y and β2 C31F mutants had a statistically significant 2-fold decrease in DHA IC₅₀ compared to the parental strain (C580Y: 12.4 nM vs. β2 C31Y: 5.2 nM and β2 C31F: 6.2 nM, p < 0.0001) (Figure 1C), mirroring the increase in sensitivity that we observed at the 0–3_hpi ring stages. This shift in IC₅₀ prompted us to investigate the sensitivity of these proteasome mutants in asynchronous cultures. Compared to Cam3.II K13^WT, Cam3.II K13^R539T, Cam3.II K13^C580Y, and Cam3.II K13^C580Y β5 A20S, which all displayed similar dose-response curves and IC₅₀ values of approximately 5 nM (Figures S1F, G), the β2 C31Y and β2 C31F mutants displayed a significant (p = 0.002) 2-fold increased sensitivity to DHA with IC₅₀ values of 2.5 nM and 2.6 nM, respectively (Figure 1D). These data show that mutations in the β2 proteasome catalytic subunit sensitizes parasites to DHA not only at ring stages, but throughout the asexual life cycle. As assessed by the RSA, β2 C31Y and β2 C31F parasites would be classified as artemisinin-resistant, since they displayed survival rates > 1%. However, relative to their Cam3.II K13^C580Y parent, these proteasome mutants display 2–3 fold lower DHA IC₅₀ values throughout their asexual lifecycles and that were below the IC₅₀ values displayed by Cam3.II K13^WT parasites, which are classified to be artemisinin-sensitive by RSA. This discrepancy highlights the subtle differences these two tests measure. The dose-response curves for trophozoite stage parasites and asynchronous parasites are shown in Figures S2B, C.

A Mutation in the β2 Proteasome Subunit Sensitizes Cam3.II K13^C580Y Parasites to OZ439

Similar to DHA, OZ439 (artefenomel) contains a peroxide bond that reacts with heme, which produces carbon-centered radicals that mediate widespread protein damage ⁴². No cross-resistance between DHA and OZ439 has been observed in parasites harboring K13^R539T or K13^C580Y in the Cam3.I and Cam3.II strains ^{38, 43}. Dose-response inhibition assays with OZ439 were conducted on 0–3_hpi ring stages, 26–30_hpi trophozoite stages, or with asynchronous parasites (Figures 2, S3, and S4). As expected, Cam3.II K13^WT, Cam3.II K13^R539T and Cam3.II K13^C580Y parasites behaved similarly in response to OZ439 at all stages tested (Figure S3) and as previously shown ³⁸. However, Cam3.II K13^C580Y β2 C31Y and β2 C31F mutants were sensitized to OZ439 compared to parental strains when tested against rings, trophozoites, or asynchronous cultures (Figure 2A–C). 0–3_hpi ring stages of the Cam3.II K13^C580Y β2 mutants were 2–4-fold more susceptible to OZ439 than the parental Cam3.II K13^C580Y parasites (C580Y: 62.7 nM vs β2 C31F: 14.8 nM (p = 0.008) and β2 C31Y: 27.1 nM (p = 0.032)) (Figure 2A). At the trophozoite stages, both the β2 C31Y and β2 C31F mutants displayed 2-fold increased sensitivity to OZ439 compared to the parental line (C580Y: 27.9 nM vs β2 C31Y: 10.8 nM and β2 C31F: 12.7 nM, p < 0.0001) (Figure 2B). This result was recapitulated when tested in asynchronous cultures (C580Y: 10.8 nM vs β2 C31Y: 4.8 nM and β2 C31F: 4.8 nM, p = 0.004) (Figure 2C). Cam3.II K13^C580Y β5 A20S displayed similar OZ439 IC₅₀ values compared to Cam3.II K13^C580Y when assayed in ring, trophozoite, or asynchronous cultures (Figure 2). Dose-response curves for parasites assayed against OZ439 are depicted in Figure S4. DHA and OZ439 IC₅₀ values determined in ring-stage, trophozoite-stage, and asynchronous parasites are summarized in Table S1. Note that K13 variants conferring DHA resistance do not display cross-resistance to OZ439 but the β2 C31Y and C31F mutations on a K13^C580Y background confer sensitivity to both peroxides.

Figure 2. Cam3.II K13^C580Y β2 proteasome mutants display increased susceptibility to OZ439.

(A) 0–3_hpi ring stages and (B) 26–30_hpi trophozoite stages isolated from Cam3.II K13^WT, Cam3.II K13^C580Y, Cam3.II K13^C580Y β2 C31Y, Cam3.II K13^C580Y β2 C31F, and Cam3.II K13^C580Y β5 A20S parasites were exposed to a range of OZ439 concentrations for 3 h, the drug was washed out, and then parasitemia was assessed approximately 66 h later. (C) Asynchronous cultures were exposed to a range of OZ439 for 72 h. Parasitemia was assessed and mean IC₅₀ values ± SEM were calculated from at least three biological replicates. Statistical significance was examined between proteasome mutants and Cam3.II K13^WT or Cam3.II K13^C580Y parasites using a Mann-Whitney U test. * p < 0.05; ** p < 0.01; **** p < 0.0001; n.s. = not significant.

Peroxides and Proteasome Inhibitors Strongly Synergize in Cam3.II K13^C580Y β5 A20S Parasites

Previously, it was shown that WLW and WLL synergize with DHA and OZ439 in Cam3.II K13^WT and Cam3.II K13^C580Y parasites at ring, trophozoite, and asynchronous stages ³⁶. To determine if the changes in susceptibility to DHA and OZ439 observed in the β2 proteasome mutants would alter the drug-drug interactions of peroxides and proteasome inhibitors, we conducted isobolograms on 26–30_hpi trophozoites isolated from proteasome mutant parasites as well as their parental strain. Parasites were exposed to fixed ratios (1:0, 4:1, 2:1, 1:1, 1:2, 1:4, 0:1) of WLW or WLL in combination with DHA or OZ439. Fractional IC₅₀ (FIC₅₀) values were calculated from these data as described ³⁶ and plotted on a Cartesian graph. Data points lying along the dotted line indicate additivity, data points below the dotted line forming a convex curve indicate synergy, and data points above the dotted line forming a concave curve indicate antagonism. As previously shown, there is synergy in Cam3.II K13^C580Y for all tested drug combinations (WLL and DHA, WLL and OZ439, WLW and DHA, and WLW and OZ439) (Figure 3, blue dots) ³⁶. Interestingly, the β5 A20S mutant displayed even greater synergy with WLL and peroxides compared to Cam3.II K13^C580Y (Figure 3A, B panels 1 vs. 4, and Figure 4), despite demonstrating increased resistance to WLL ³⁶, confirmed in (Figure S5A). The susceptibility of Cam3.II K13^C580Y β2 C31F to WLL and WLW was previously unknown. Here, we determined that the β2 C31F mutant, like the previously tested β2 C31Y mutant, exhibited increased sensitivity to WLL (Figure S5A). Cam3.II K13^C580Y β2 C31F was more resistant to WLW than its parent Cam3.II K13^C580Y but not as resistant as Cam3.II K13^C580Y β2 C31Y (Figure S5B). Both β2 mutants demonstrated additive effects when exposed to WLL and DHA, WLL and OZ439, or WLW and DHA (Figure 3A–C). However, WLW and OZ439 retained synergistic activities in the β2 mutants (Figure 3D). A global overview of these drug-drug interactions is shown in Figure 4. Together, these data demonstrate that in the unlikely event that parasite resistance arises to WLL or WLW, the most common mutations will not compromise susceptibility to DHA or drug-drug interactions. In some instances, proteasome subunit mutations will confer increased susceptibility to DHA and OZ439, or maintain synergistic interactions between peroxides and proteasome inhibitors.

Figure 3. Peroxides and proteasome inhibitors are strongly synergistic in Cam3.II K13^C580Y β5 A20S parasites.

Isobolograms were conducted on 26–30_hpi trophozoites isolated from Cam3.II K13^C580Y, Cam3.II K13^C580Y β2 C31Y, Cam3.II K13^C580Y β2 C31F, and Cam3.II K13^C580Y β5 A20S strains. Parasites were exposed to a range of (A) WLL and DHA, (B) WLL and OZ439, (C) WLW and DHA, or (D) WLW and OZ439 at fixed ratios (1:0, 4:1, 2:1, 1:1, 1:2, 1:4, 0:1) for 3 h, and then the drugs were washed out and parasitemia was assessed in the following replication cycle. Fractional IC₅₀ (FIC₅₀) values from 2 independent experiments (indicated by different shading) were calculated and plotted on a Cartesian graph. The diagonal dotted black line represents the isobole line which represents perfect additivity. Data points along this line indicate additivity and data points that fall below the isobole line along a convex curve indicate synergy. Cam3.II K13^C580Y, Cam3.II K13^C580Y β2 C31Y, Cam3.II K13^C580Y β2 C31F, and Cam3.II K13^C580Y β5 A20S are indicated in blue, pink, orange, and green, respectively.

Figure 4. Proteasome inhibitors and peroxides are synergistic or additive in Cam3.II K13^C580Y β2 and β5 proteasome mutants.

The heatmap represents mean ∑FIC₅₀ values derived from isobologram analyses performed on 26–30_hpi trophozoite stages isolated from the indicated parasites. Values obtained range from 0.57, indicating synergy in blue, to values of 1.01, indicating additivity in white. The following drug combinations were tested: WLL and DHA, WLL and OZ439, WLW and DHA, and WLW and OZ439. Mean ∑FIC₅₀ values close to 0 indicates synergy, values close to 1 indicates additivity, while values close to 2 indicate antagonism.

Discussion

The Plasmodium-specific vinyl sulfone proteasome inhibitors WLL and WLW potently inhibit both artemisinin-sensitive and artemisinin-resistant parasites and demonstrate strong synergy with DHA and OZ439 ³⁶. Thus, proteasome inhibitors are attractive partner drugs to overcome current problems with artemisinin resistance. In vitro, resistance to WLL and WLW was difficult to generate, with the majority of resistance-associated mutations observed in the β2 and β5 catalytic subunits ³⁶. Here, we show that the most frequently recovered WLW-selected mutations, β2 C31Y and β2 C31F, sensitize Cam3.II K13^C580Y parasites to DHA and OZ439 when exposed to these drugs at the early ring stages, mid-trophozoite stages, and in asynchronous cultures. In fact, these parasites displayed approximately 2-fold higher sensitivity to DHA and OZ439 than artemisinin-sensitive Cam3.II K13^WT parasites. The singular change of K13^WT to K13^C580Y confers artemisinin resistance in the Cam3.II strain ¹⁴ but does not confer resistance to the related peroxide OZ439 ^{38, 43}. Yet we observe that an additional mutation at the β2 locus ³⁶ sensitizes Cam3.II K13^C580Y parasites to both DHA and OZ439 at early ring stages, trophozoite stages and in asynchronous parasites. This demonstrates an uncoupling between the contribution of K13 and proteasomes to differential peroxide susceptibility and underscores the developmental stage specificity of each peroxide. Our data also show that it is difficult to predict drug-drug interactions. For example, only the WLW and OZ439 drug combination displayed synergy in Cam3.II K13^C580Y β2 mutants, which had increased susceptibility to both DHA and OZ439. In contrast, all tested combinations of peroxides and proteasome inhibitors demonstrated synergy in Cam3.II K13^C580Y β5 A20S parasites, which have a 4.5-fold increased resistance to WLL, a 2-fold decreased resistance to WLW, and no difference in susceptibilities to DHA and OZ439 when compared to its parent Cam3.II K13^C580Y. This demonstrates the unpredictability of drug-drug interactions and necessity for empirical determination of emergent combinatorial drug effects.

K13^PDmut are believed to mediate resistance by decreasing hemoglobin uptake in early ring stages, thus reducing peroxide activation ^17–19. In contrast, proteasomes are important throughout the parasite lifecycle ^{28, 29} and are upregulated during the mature stages ^{40, 41}. Recently, proteasomes have been shown to be secreted in extracellular vesicles delivered to naive RBC, where this proteolytic complex degrades cytoskeleton proteins, thus modifying the host RBC to promote parasite invasion ⁴⁴. While differential K13-mediated artemisinin drug responses are not observed outside of early ring stages, β2 proteasome mutants show a marked decrease in sensitivity to DHA and OZ439 compared to the parental Cam3.II K13^C580Y throughout the asexual life cycle. There are a number of reasons that can explain these phenomena. After heme-mediated activation, DHA and OZ439 are known to nonspecifically alkylate heme and parasite proteins in the immediate vicinity ^{20, 21, 42, 45}. Presumably, alkylation of parasite proteins impairs protein function, thus impairing parasite viability. These nonfunctional proteins are recognized by chaperones as misfolded or damaged proteins which will then be ubiquitinated and targeted for proteasome-mediated degradation. Whether K13 acts as an E3 ubiquitin ligase substrate adaptor in this cellular quality control process remains to be seen. DHA has been shown to inhibit proteolytic activity of the β5 subunit, although only at about 50% of the β5-targeting proteasome inhibitor epoxomicin ⁴⁶. Whether DHA inhibits the other catalytic subunits is unknown. Our data implicate the importance of the proteasome in artemisinin and OZ439 response. Whether the β2 C31Y and C31F mutations directly impair proteolytic activity or if these mutations allosterically regulate proteasome activity, and if these changes in activity impact proteasome-mediated RBC remodelling remains to be explored. Although the proteasome possesses three catalytic subunits, most of the proteolytic activity is attributed to the β5 subunit. One would surmise that it would be difficult to obtain a viable parasite with a mutation in the main catalytic subunit that compromised its ability to degrade proteins, given the necessity of the proteasome throughout the parasite lifecycle ^{28, 29, 44}. Thus, it is perhaps unsurprising that the β5 mutant obtained in our in vitro drug selections does not sensitize Cam3.II K13^C580Y to peroxides, if this sensitization is due to loss of proteolytic activity.

The data presented here provide compelling evidence toward the future of proteasome inhibitor-peroxide combinations to combat artemisinin resistance. Should these combination therapies advance to clinical use, we will likely not see resistant parasites arise because: (1) WLL and WLW are equally potent on artemisinin-sensitive and artemisinin-resistant parasites, (2) it was difficult to generate resistance against WLL and WLW in vitro, (3) the majority of the mutations which mediate resistance to WLW (β2 mutants) will be selected against due to their increased sensitivity to peroxides, (4) the WLL-selected β5 mutation did not compromise DHA and OZ439 potency, and (5) peroxides and proteasome inhibitors were highly synergistic in the β5 mutant. Mutations in other proteasome subunits and on different parasite backgrounds have yet to be investigated, and it remains to be seen if these results will be recapitulated in vivo.

Methods

Parasite Culturing

Cam3.II K13^WT, Cam3.II K13^R539T (RF967), and Cam3.II K13^C580Y are laboratory-adapted Cambodian isolates that were genetically engineered at the K13 locus as described in the literature ¹⁴. Cam3.II K13^C580Y parasites expressing β2 C31Y, β2 C31F, or β5 A20S were obtained from in vitro selection studies with WLL and WLW as described in the literature ³⁶. Parasite cultures were propagated in O+ red blood cells (RBCs) (anonymous donors, purchased from Interstate Blood Bank, Memphis, Tennessee). RBCs were stored at 50% hematocrit in ADSOL (2 mM adenine (Alfa Aesar, Haverhill, Massachusetts), 111 mM dextrose (Fisher BioReagents, Pittsburgh, Pennsylvania), 41.2 mM mannitol (Acros Organics, Fair Lawn, New Jersey), and 154 mM sodium chloride (Fisher BioReagents) for prolonged RBC vitality as described in ref ⁴⁷ at 4^oC. Parasites were cultured in complete media (RPMI 1640 media supplemented with 0.01 mg/mL gentamicin (Gibco, Dun Laoghaire, Co Dublin, Ireland), 50 mg/mL hypoxanthine (Acros Organics), and 0.5% Albumax II (Invitrogen, Carlsbad, California)) at 5% hematocrit, and were maintained at 37^oC in a Heracell™ VIOS 160i CO₂ Incubator (Thermo Fisher Scientific, Waltham, Massachusetts) under hypoxic conditions (5% O₂, 5% CO₂, 90% N₂). Gas was purchased from Matheson Gas (Irving, Texas).

Stage Synchronization

0–3_hpi ring stages were obtained as previously described ³⁷ with minor modifications. Briefly, cultures were treated with 5% sorbitol (Acros Organics) at 37 ^oC for 15 min and then cultured for 33 h. Schizont-enriched cultures were resuspended in RPMI 1640 supplemented with 14.3 U/mL sodium heparin (Merck, Darmstadt, Germany) and incubated at 37 ^oC for 30 minutes with intermittent vortexing. Schizont stages were purified using a 75% Percoll (GE Healthcare) density gradient followed by centrifuging at 2200g for 15 min. Purified schizont stages were washed once with RPMI 1640 supplemented with 14.3 U/mL sodium heparin, then propagated in fresh RBCs and complete media at 2% hematocrit. Parasite cultures were incubated for 3 h, and any remaining late-stage parasites were removed with 5% sorbitol, leaving 0–3_hpi rings. Trophozoite stage parasites (26–30_hpi) were obtained by exposing cultures to two consecutive treatments of 5% sorbitol, 12 h apart. Parasites were then cultured for an additional 12 h. Stage isolation was verified using Giemsa-stained thin blood smears and imaged via light microscopy using a 100× oil immersion objective.

Growth Inhibition and Survival Assays

Parasites were synchronized as described above, and then seeded in 96-well plates (Thermo Fisher Scientific) at 1% hematocrit and 0.7% parasitemia for ring stage survival assays (RSAs) and ring-stage growth inhibition assays, or 0.2% parasitemia for trophozoite stage and asynchronous growth inhibition assays in a total of 200 μL per well. Custom drug concentrations were dispensed with a MANTIS^® microfluidic liquid handler (FORMULATRIX^®, Bedford, Massachusetts) or were dispensed by hand in 2-fold serial dilutions. Ring and trophozoite stage parasites were treated for 3 h in U-bottom plates, then drug was washed out by centrifuging 96-well plates at 1500 rpm for 1 min, removing media, and resuspending in 190 μL of complete media, which removed >95% of medium per wash ^{38, 48}. Following four washes, cultures were transferred to a new flat-bottom 96-well plate (Thermo Fisher Scientific) ^{38, 48} and cultured for an additional 66 h. Asynchronous parasites were treated for 72 h in flat-bottom plates.

Parasites were measured by staining parasite nuclei with 2 μg/mL Hoechst 33342 (Thermo Fisher Scientific) and respirating mitochondria with 100 nM MitoTracker Deep Red FM (Thermo Fisher Scientific). CellCarrier-96 Ultra Microplates (PerkinElmer, Waltham, MA) were coated with 0.1 mg/mL poly-L-lysine (MP Biomedicals, Irvine, CA). 40 μL of stain was added per well. Then, 10 μL of parasite culture at 1% hematocrit, approximately 10 million RBC, was added per well. Cells were allowed to settle for 1 h at 37 ^oC under normal culturing conditions as described above. Plates were washed twice with 1 × PBS, taking care to not disturb RBCs. Cells were fixed by adding 50 μL of 2% paraformaldehyde (Alfa Aesar) and 2% glutaraldehyde (Fisher Scientific) to each well and incubating at room temperature for 30 min. Following fixation, the fixative was removed and 150 μL 1 × PBS was added to each well. Plates were imaged with an Operetta CLS High-Content Imaging System (PerkinElmer) using the following settings: 20× air objective, nonconfocal, binning 2. Channels were captured in the following order: Hoechst (excitation/emission (ex/em) 355–385/430–500), Mitotracker Deep Red (ex/em 615–645/655–760), and brightfield. For each channel, focus height was determined, and exposure time and percent power were adjusted so that the intensity for all channels was approximately 4000–5000 counts/pixel. At least 10,000 cells were imaged per sample. Images were analyzed using Harmony 4.9 software with PhenoLOGIC. GraphPad Prism software version 9 was used to calculate IC₅₀ values by nonlinear regression analysis. Survival in RSAs were calculated by dividing the parasitemia of cultures treated at 700nM by the parasitemia of DMSO-treated cultures. Statistical analyses were performed using Mann-Whitney U tests. At least three biological replicates were performed for each assay.

Isobolograms

Trophozoite stages were isolated as described above and then seeded in round-bottom 96-well plates at 1% hematocrit and 0.2% parasitemia. Starting concentrations of drug were prepared in 1:0, 4:1, 2:1, 1:1, 1:2, 1:4 ratios and 2-fold serial dilutions were dispensed by hand or using a MANTIS microfluidic liquid handler. Drug incubation, washout, and analysis were performed as described above. FIC₅₀ values were calculated by dividing the IC₅₀ value obtained when exposed to both compounds by the IC₅₀ obtained when exposed to one compound alone.

$${\text{FIC}}_{50}\text{=}\frac{\text{IC}50\text{in}\text{presence}\text{of}\text{compounds}\text{A}\text{and}\text{B}}{\text{IC}50\text{in}\text{presence}\text{of}\text{compound}\text{A}}$$

FIC₅₀ values are plotted on a Cartesian graph in Figure 3. Mean ∑FIC₅₀ was calculated by averaging the sums of FIC₅₀ values at each combination and plotted as a heat map in Figure 4. Isobolograms and heat maps were plotted using GraphPad Prism 9 software.

Drug Compounds

DHA was purchased from Sigma-Aldrich (St. Louis, MO). OZ439 was kindly provided by Dr. J. L. Vennerstrom (University of Nebraska Medical Center). WLL and WLW were kindly provided by Dr. M. Bogyo (Stanford School of Medicine). All compounds were solubilized by dissolving in DMSO (Fisher Scientific), and maximum DMSO concentrations in all assays were less than 0.2%.

Supplementary Material

Supporting Information

Figure S1. DHA dose-response inhibition assays conducted on K13 isogenic Cam3.II parasites.

Figure S2. Dose-response curves show increased susceptibility of Cam3.II K13^C580Y β2 proteasome mutants to DHA.

Figure S3. OZ439 dose-response inhibition assays conducted on K13 isogenic Cam3.II parasites.

Figure S4. Dose-response curves show increased susceptibility of Cam3.II K13^C580Y β2 proteasome mutants to OZ439.

Figure S5. Susceptibility of Cam3.II K13^C580Y parent and the β2 and β5 proteasome mutants to WLL and WLW.

Table S1. Summary of DHA and OZ439 IC50 values.

Abbreviations

DHA

dihydroartemisinin

FIC₅₀

Fractional IC₅₀

hpi

hours post invasion

RSA

ring-stage survival assay