Impact of field isolate identified non-synonymous single nucleotide polymorphisms on Plasmodium falciparum Equilibrative Nucleoside Transporter 1 (PfENT1) inhibitor efficacy

Abstract

Plasmodium falciparum causes the most severe form of malaria and causes approximately 500,000 deaths per year. P. falciparum parasites resistant to current anti-malarial treatments are spreading. Therefore, it is imperative to develop new anti-malarial drugs. Malaria parasites are purine auxotrophic. They rely on purine import from the host erythrocyte via Equilibrative Nucleoside Transporters (ENTs). Recently, inhibitors of the P. falciparum ENT1 (PfENT1) that inhibit proliferation of malaria parasites in culture have been identified as promising starting points for antimalarial drug development. Genome sequencing of P. falciparum field isolates has identified non-synonymous single nucleotide polymorphisms (SNPs) in the gene encoding PfENT1. Here we evaluate the impact of these PfENT1 SNPs on purine substrate affinity and inhibitor efficacy. We expressed each PfENT1-SNP in Saccharomyces cerevisiae. Using PfENT1-SNP-expressing yeast, we characterized the PfENT1 purine substrate affinity using radiolabeled substrate uptake inhibition experiments. Four of the 13 SNPs altered affinity for one or more purines by up to 7-fold. Three of the SNPs reduced the potency of a subset of the inhibitors by up to 7-fold. One SNP, Q284E, reduced the potency of all six inhibitor chemotypes. We tested drug efficacy in available parasite strains containing PfENT1 SNPs. While PfENT1-SNP-expressing yeast had decreased sensitivity to PfENT1 inhibitors, parasite strains containing SNPs showed similar or more potent inhibition of proliferation with all PfENT1 inhibitors. Thus, parasite strains bearing PfENT1 SNPs are not resistant to these PfENT1 inhibitors. This supports PfENT1 as a promising target for further development of novel antimalarial drugs.

Graphical Abstract

Malaria is a major public health issue; in 2016 there were an estimated 216 million new malaria cases and 445,000 deaths due to malaria worldwide.¹ Young children and pregnant women in sub-Saharan Africa are most likely to die of malaria. Malaria is caused by infection with parasites from the genus Plasmodium transmitted to humans through mosquito vectors. Infection with Plasmodium falciparum causes the most severe form of malaria and is associated with the highest mortality. Global trends in malaria incidence and mortality have shown improvements, however, there is increasing evidence that P. falciparum parasites resistant to antimalarial drugs are spreading. Resistance to new anti-malarial drugs could occur as a result of preexisting genetic diversity and mutations in a parasite population. Naturally occurring non-synonymous single nucleotide polymorphisms (SNPs) in target proteins could give parasites an increased fitness advantage against new drugs. SNPs in target proteins have already been shown to confer resistance to many anti-malarials.^2,3,4 Therefore, it is important to determine whether known naturally occurring P. falciparum SNPs in the drug target protein decrease the potency of potentially new antimalarial drugs.

The current recommended antimalarial treatment is artemisinin-based combination therapy (ACT). ACT is a combination of an artemisinin derivative with a longer acting antimalarial that has a different mode of action. Progress in malaria control can be partly attributed to the efficacy of ACT. However, in as little as a decade after the introduction of ACT, artemisinin resistance has been confirmed in five countries in the Greater Mekong subregion (GMS).¹ Recent reports show P. falciparum strains with reduced response to artemisinin and treatment failure rates up to 26% in Vietnam as well as resistance to the latest recommended ACT therapy, dihydroartemisinin (DHA)-piperaquine (PP) in Western Cambodia.^5,6,7 The spread of ACT resistance may lead to millions of deaths. Therefore, it is imperative to identify new drug targets for antimalarial drug development.

Plasmodium parasites are vulnerable to inhibitors of purine transport and the purine salvage pathway because they are purine auxotrophs. Previously, the Plasmodium purine salvage pathway was identified as a potential target for the development of anti-malarial drugs.^8,9,10,11 The equilibrative nucleoside transporter, PfENT1, is the primary pathway for purine uptake into the parasite. PfENT1-knockout parasites (pfent1Δ) are not viable in growth media containing physiological purine levels found in human blood (<10 μM hypoxanthine).⁹ PfENT1-knockout parasites can only survive when grown with supra-physiologic purine concentrations (>50 μM hypoxanthine).^9,12,11 PfENT1-knockout parasites must therefore have an as yet unidentified secondary purine transport pathway. A recent study used random retrotransposon insertion mutagenesis to identify P. falciparum genes essential for blood stage parasite proliferation. They categorized PfENT1 as an essential gene.¹³ These results demonstrate that PfENT1 is essential for blood stage P. falciparum parasite proliferation. They support the hypothesis that PfENT1 is a potential drug target for antimalarial drug development.

Previously, using a yeast-based high-throughput screen (HTS) we identified inhibitors of the PfENT1 transporter.¹⁴ We identified 171 hits from a screen of 64,500 compounds. Nine of the highest activity compounds with six distinct chemical scaffolds were extensively characterized (Table 1). These compounds inhibited [³H]adenosine uptake into PfENT1-expressing yeast and into erythrocyte-free trophozoite stage parasites with IC₅₀ values in the 5–50 nM range. When tested in chloroquine -sensitive and –resistant P. falciparum parasites the compounds killed parasites with 5–50 μM IC₅₀ values.¹⁴ P. vivax infection has significant geographic overlap with P. falciparum and is associated with high morbidity.¹⁵ All nine PfENT1 inhibitors blocked [³H]adenosine uptake in PvENT1-expressing yeast with similar efficacy.¹⁶ Additionally, when inhibitor efficacy was tested in yeast expressing five known PvENT1 SNPs the inhibitor potency was minimally impacted.¹⁶ These results strongly support the ability to target both falciparum and vivax ENT1 for the development of novel antimalarial drugs.

Table 1.

Structures of nine PfENT1 inhibitors identified in a yeast-based high throughput screen (HTS) described previously in Frame et al., 2015.¹⁴ To maintain consistency with our previous work, the structures of the nine compounds are labeled by their rank order number based on potency in the initial HTS screen. They are grouped into six distinct chemotypes (A-F) based on chemical scaffold similarities.

Chemotype	Compound Structure
A
B
C
D
E
F

Although PfENT1 is a promising target for antimalarial drug development it is important to evaluate whether the existence of naturally occurring PfENT1 SNPs has an effect on purine uptake and PfENT1 inhibitor efficacy. Existing SNPs could cause resistance to our inhibitors and limit their potential for drug development. We searched the PlasmoDB database (http://www.plasmodb.org/plasmo/) for P. falciparum pfent1 gene sequences in field isolates. We used the 3D7 strain sequence as the wild type (WT) and defined non-synonymous SNPs as changes from the 3D7 sequence. We identified 13 naturally occurring non-synonymous SNPs. Two of these SNPs are found in laboratory parasite strains, HB3 and 7G8. In this study we evaluated whether the PfENT1 SNPs have effects on purine uptake and/or PfENT1 inhibitor efficacy.

We found that most PfENT1-SNPs did not cause a significant change in the purine nucleobase and nucleoside affinity. The potency of PfENT1 inhibitors to inhibit radiolabeled substrate uptake into yeast expressing two of the SNPs found in parasite field isolates, Q284E and F36S, was decreased when compared to WT PfENT1-expressing yeast. However, the potency of the PfENT1 inhibitors with the laboratory strains 7G8, which contains the PfENT1 Q284E SNP, and HB3, which contains the PfENT1 F36S/V129I SNP, in parasite viability assays was either unchanged or increased when compared to the WT 3D7 P. falciparum strain.

Results and Discussion

Characterization of PfENT1-SNP purine uptake and affinity

We hypothesize that the existence of non-synonymous SNPs in PfENT1 could cause resistance to some PfENT1 inhibitors. The PlasmoDB database contains whole genome sequences for about 100 laboratory and field-isolated P. falciparum strains. We searched the pfent1 gene sequence and identified 13 naturally occurring non-synonymous SNPs. The positions of all of the SNPs are shown on a transmembrane topology plot (Figure 1). Several mutations occur as single mutations in PfENT1. These include F36S, F92I, L197F, Q284E, S317T, V345I, A351S, L385I, F394L, I418M. Some of these mutations also occur in combination with one or more additional mutations. These include F36S/V129I, V129I/L133F, and F36S/V129I/L133F/P417L (subsequently referred to as the “Quad” SNP). P417L does not occur alone. V129I is also only seen coupled with either F36S or L133F. L133F never occurs alone and is present in the Quad mutant or with V129I. We included these combinations of mutations in our studies.

Figure 1.

Transmembrane topology plot for PfENT1 showing the putative locations of the 13 SNPs.

Previously, our lab generated a yeast-codon-optimized pfent1 gene in the pM189 episomal yeast expression plasmid to functionally express Plasmodium ENT1 homologues in yeast for transporter characterization and identification of transporter inhibitors.^14,16,17 We generated constructs containing each of the 13 SNPs in this plasmid. We expressed these constructs in ade2Δ purine auxotrophic yeast that also lacked the endogenous uridine transporter, FUI1 (fui1Δ). The purine auxotrophic yeast can grow with adenine as the sole purine source because adenine can be transported through the endogenous yeast FCY2 transporter. Purine auxotrophic yeast cannot grow when adenosine is the sole purine source because yeast lack an endogenous adenosine transporter. However, PfENT1 can transport adenosine. So when they express a functional Plasmodium ENT1 homologue, purine auxotrophic yeast are able to grow when adenosine is the sole purine source. All 13 PfENT1 SNPs were expressed in yeast and grew in media containing adenosine as the sole purine source. This demonstrated the functional expression of the SNP containing transporters.

We hypothesize that the PfENT1 SNPs may alter the purine affinity compared to WT. To characterize PfENT1-SNP purine affinity, we tested the ability of increasing concentrations of unlabeled purines to inhibit uptake of [³H]uridine (Figure 2). PfENT1 transports uridine, which does not share downstream metabolic enzymes with the purines tested. Therefore, we can assume that competition between the purine and the [³H]uridine can only occur at the transporter.^18,19 PfENT1 mediated [³H]uridine uptake was linear over a 30 min time interval (Supplementary Figure S1). Uptake experiments lasted 15 min and were in the linear uptake phase. We tested two purine nucleobases, hypoxanthine and adenine, as well as their corresponding purine nucleosides, inosine and adenosine. Full inhibition curves were obtained for all 13 PfENT1-SNP expressing yeast. Figure 2 illustrates experiments for the concentration-dependent inhibition of [³H]uridine uptake by the indicated purines into yeast expressing and WT PfENT1 or the F36S/V129I or Q284E SNPs. For all 13 SNPs and WT, we quantified the potency of each purine by the concentration that caused 50% inhibition of [³H]uridine uptake (IC₅₀). Figure 3 shows the ratio of purine IC₅₀ values for the SNPs compared to WT-PfENT1. Nine SNPs had IC₅₀ values that were not significantly different than WT for any of the purines tested (Figure 3, Table 2). However, three mutants, Q284E, A351S, and F394L, had significantly increased IC₅₀ values for one or both nucleobases. The hypoxanthine IC₅₀ value was 3-fold higher for Q284E (p ≤ 0.05). The adenine IC₅₀ value was 2-fold higher for A351S (p ≤ 0.001). For F394L the IC₅₀ values for both hypoxanthine and adenine were increased 2- to 4-fold (p ≤ 0.0001). The only SNP with significantly lower IC₅₀ value for a nucleobase was L197F where the adenine IC₅₀ value was 13-fold lower (p ≤ 0.05). For all SNPs tested, adenosine IC₅₀ values were not significantly different from WT (Figure 3, Table 2). Thus, most of the SNPs had limited effects on the affinity of PfENT1 for these nucleobases and nucleosides.

Figure 2.

Concentration–dependent inhibition of [³H]uridine uptake by the indicated purine into yeast expressing either WT or SNP containing PfENT1. Results show the average of n≥3 biological replicate experiments. Mean ± SEM is shown. Table 2 contains the average IC₅₀ values for each purine and SNP.

Figure 3.

Ratio of IC₅₀ values (n=3) for inhibition of [³H]uridine uptake into PfENT1 SNP-expressing yeast by the indicated purine compared to the IC₅₀ value in wild-type (WT) PfENT1 expressing yeast. SNP containing F36S/V129I/L133F/P417L is named “Quad”. Dashed line indicates a ratio of 1 and no difference in the IC₅₀ values. Symbols represent the values of individual experiments. Bars show mean ± SEM. Significantly different than WT by one-way ANOVA *(p ≤ 0.05), **(p ≤ 0.01), ***(p ≤ 0.001), ****(p ≤ 0.0001).

Table 2.

Purine IC₅₀ values for inhibition of [³H]uridine uptake by purine nucleobases and nucleosides in yeast expressing PfENT1 mutations identified from P. falciparum field isolates. SNP containing F36S/V129I/L133F/P417L is named “Quad”. Average IC₅₀ values are from three independent trials.

IC50 values for inhibition of [3H]uridine uptake by indicated purine (μM)
SNP	Hypoxanthine	Inosine	Adenine	Adenosine
WT	178 ± 44	20 ± 7	475 ± 124	139 ± 21
F36S	171 ± 25	24 ± 7	292 ± 21	158 ± 34
F92I	207 ± 34	51 ± 14	176 ± 35	276 ± 71
L197F	282 ± 8	37 ± 12	46 ± 5*	79 ± 33
Q284E	566 ± 133*	18 ± 6	1068 ± 128**	140 ± 93
S317T	175 ± 31	28 ± 2	205 ± 49	178 ± 14
V345I	161 ± 57	4 ± 0.2	225 ± 17	36 ± 2
A351S	576 ± 270*	26 ± 9	1242 ± 142***	184 ± 28
L385I	173 ± 46	58 ± 17	185 ± 15	248 ± 65
F394L	837 ± 93****	94 ± 29	1498 ± 227****	162 ± 62
I418M	145 ± 7	12 ± 2	221 ± 36	97 ± 3
F36S/V129I	107 ± 5	14 ± 2	333 ± 32	98 ± 4
V129I/L133F	360 ± 66	20 ± 7	930 ± 285	206 ± 49
Quad	108 ± 5	12 ± 1	362 ± 56	112 ± 5

Data show mean ± SEM. Significantly different than WT by one-way ANOVA,

^*(p ≤ 0.05),

^**(p ≤ 0.01),

^***(p ≤ 0.001),

^****(p ≤ 0.0001).

Riegelhaupt et al., 2010 investigated the functional effects of the F36S, L133F, and F394L polymorphisms using the Xenopus laevis oocyte expression system.²⁰ The kinetics of [³H]adenosine and [³H]hypoxanthine uptake were evaluated to determine the K_m for the expressed SNPs. Riegelhaupt et al., 2010 found that mutants F36S and L133F did not have any effect on the adenosine K_m. In our study, F36S alone does not have any significant effect on the IC₅₀ values of either nucleobases or nucleosides compared to WT. We also find that F36S and L133F combined with V129I have no significant effect on the IC₅₀ values of both nucleobases and nucleosides. Riegelhaupt et al., also reported a 3-fold decrease in K_m for both adenosine and hypoxanthine for F394L when compared to WT PfENT1. We find that the F394L has no significant effect on nucleoside IC₅₀ values. However, F394L causes a 3- to 5–fold increase in nucleobase IC₅₀ values compared to WT (p ≤ 0.001). We have no obvious explanation for the difference between the results for F394L in our current experiments and those performed with the Xenopus laevis expression system. Possible differences include differences in lipid composition, e.g. oocytes contain cholesterol whereas yeast have ergosterol, or differences in phosphorylation or glycosylation.^21,22,23

F394L is located within a highly conserved GXXXG motif found in TM11 of all ENT homologues (Figure 1). Based on substituted cysteine accessibility experiments, Riegelhaupt et al. suggested that, TM11 lined part of the PfENT1 purine permeation pathway.²⁰ Introduction of cysteine at each the conserved glycine residues in the GXXXG motif (Gly393, Gly396) abolished adenosine uptake and significantly reduced PfENT1 protein expression levels. In other proteins GXXXG motifs have been shown to play an important role in protein folding, dimerization, and structural integrity.^24,25,26,27 Substitution of a cysteine at position 394 is not reactive with sulfhydryl-reactive methanethiosulfonate reagents. This suggests that the position may be buried in the interior of PfENT1 and might interact with another transmembrane segment or the lipid bilayer.²⁰ Thus, any effects of mutations at position 394 are probably indirect and mediated by changes in protein conformation that ultimately affect residues that interact with transported substrates.

TM2 and TM10 lie adjacent to TM11 and may play a role in PfENT1 permeation. Residues within TM10 were identified as at least transiently on the water-accessible transporter surface.²⁸ Among these residues, A351 was identified as water-accessible, suggesting it may be part of the PfENT1 permeation pathway (Figure 1). Nothing is known about the water surface accessibility of the other SNP mutants that significantly changed affinity for purine transport.

Effect of PfENT1 non-synonymous SNPs on sensitivity to PfENT1 inhibitors

The nine PfENT1 inhibitors used in this study are identified by the rank order number of their effect in the yeast-based HTS screen of 64,500 compounds. The nine PfENT1 inhibitors include the first compound identified in the HTS (compound 19), the top seven hits, and the next compound identified with a distinct chemotype in the HTS ranking (compound 13) (Table 1). The nine compounds represent six distinct chemotypes (chemotype A-F) (Table 1). Four of the compounds have a common chemical scaffold and are grouped into one chemotype (chemotype B) (Table 1).

We tested the ability of the nine compounds to inhibit [³H]adenosine uptake into each of the PfENT1-SNP-expressing yeast. Figure 4 shows average results for the concentration-dependent inhibition of [³H]adenosine uptake by PfENT1 inhibitors in yeast expressing WT and four of the SNPs. All compounds inhibited [³H]adenosine uptake in a concentration dependent manner and had IC₅₀ values within a range of 2 to 244 nM. Figure 5 shows the ratio of IC₅₀ values for inhibition of [³H]adenosine uptake by PfENT1 inhibitors in PfENT1-SNP expressing yeast compared to the corresponding IC₅₀ values in yeast expressing WT-PfENT1. A ratio value less than one indicates an increased sensitivity to inhibition by compounds, while a ratio greater than one indicates decreased sensitivity to inhibition by compounds. Due to limited availability of compound 1, the SNPs L197F, S317T, and L385I were not tested with chemotype A.

Figure 4.

Concentration-dependent inhibition of [³H]adenosine uptake by PfENT1 inhibiting compounds into yeast expressing WT or SNP containing PfENT1 by the indicated compounds. Compound 7 was used for chemotype B. Results show the average of n≥3 biological replicate experiments. Mean ± SEM is shown. Error bars are smaller than the symbols for some points.

Figure 5.

Ratio of IC₅₀ values (n=3) for inhibition of [³H]adenosine uptake in PfENT1 SNP-expressing yeast by PfENT1 inhibiting compounds compared to the IC₅₀ value in WT PfENT1-expressing yeast. SNP containing F36S/V129I/L133F/P417L is named “Quad”. Dashed line indicates a ratio of 1 and no difference in the IC₅₀ values. Symbols represent the values of individual experiments. Bars show mean ± SEM. Significantly different than WT by one-way ANOVA *(p ≤ 0.05), **(p ≤ 0.01), ***(p ≤ 0.001), ****(p ≤ 0.0001). Compound numbers as per Frame et al., (2015).

The effect of PfENT1-SNPs on the compound IC₅₀ values varied across and within all chemotypes tested. For the majority of the SNPs tested, the potency of the inhibitors was not significantly different compared to their potency with WT (Figure 5). The Q284E mutation reduced the inhibitory potency of all six chemotypes. With the exception of compound 19, Q284E caused a 3- to 7-fold decrease in sensitivity for all nine PfENT1 inhibitors (p ≤ 0.01) (Figures 4 and 5). The F36S mutation caused a 4-fold to 6-fold decrease in sensitivity to compounds 7 and 19 in chemotype B (p ≤ 0.001) and a 3-fold (p ≤ 0.05) decrease in the potency of chemotype E. The Quad SNP reduced the potency of compound 19 in chemotype B by 2-fold (p ≤ 0.05).

These results suggest that the binding of all nine compounds was sensitive to the amino acid at position 284 in transmembrane segment 8. Position 284 may be part of a common binding site for all nine compounds or the specific amino acid at this position may cause a conformational change that affects distant compound-binding site(s). Although the Q284E mutation results in a negative charge at this position, it is unlikely that any effects caused by the mutation are due to a direct electrostatic effect because none of the nine compounds are charged.

While SNP F36S was less sensitive to chemotypes B and E, when combined with V129I (SNP F36S/V129I) the sensitivity was unchanged. The Quad SNP, which also includes the F36S mutation, is less sensitive to both chemotype B and E although the magnitude of the effect is smaller than in F36S alone. The effect of the Quad mutant is only significant for compound 19 in chemotype B (p ≤ 0.05) (Figure 5).

PfENT1 inhibitor efficacy on parasite lines containing PfENT1 SNPs

Next, we tested the ability of PfENT1 inhibitors to inhibit proliferation of several drug sensitive and drug resistant parasite strains using 48 hour parasite growth assays.^29,30,31 The strains we tested include 3D7, HB3, 7G8, and Cam3.II (ART)^R. The 3D7 strain originates from Africa and is what we consider WT. 3D7 is chloroquine sensitive and sulfadoxine resistant. The HB3 strain originates from Honduras and contains the F36S/V129I SNPs. HB3 is sensitive to chloroquine and resistant to sulfadoxine. The 7G8 strain originates from Brazil and contains the Q284E mutation. 7G8 is resistant to both chloroquine and pyrimethamine. The Cam3.II (ART)^R originates from Southeast Asia and is artemisinin-resistant.

We found that for these three parasite lines, the IC₅₀ values for all nine PfENT1 inhibitors were similar to or lower than the IC₅₀ values for the 3D7 WT parasite strain (Figures 6 and 7, Table 3). Figure 7 shows the ratio of PfENT1 inhibitor IC₅₀ values for inhibition of parasite strain proliferation as compared to IC₅₀ values for inhibition of WT 3D7 parasites. Ratio values greater than 1 indicate decreased drug potency against parasites while ratio values less than 1 indicate an increase in drug potency to inhibit proliferation of parasites. All parasite strains tested had ratio values less than 1 for all nine PfENT1 inhibitors (Figure 7). PfENT1 inhibitors were up to 10-fold more potent against these parasite strains compared to WT (Table 3). Surprisingly, there was no correlation between the effects of SNPs in the yeast based assay and in the parasite proliferation assay. In the yeast expression system, the Q284E mutation reduced inhibitor potency but the inhibitor potency was increased in the 7G8 parasite strain that contains the Q284E mutation. There are many potential explanations for the differential effects on inhibitor IC₅₀ values. It may be due to differences in yeast and parasites that might include 1) different plasma membrane lipid composition and 2) potentially different phosphorylation and glycosylation. The assays are also quite different, the yeast assay is a 15 minute [³H]adenosine uptake assay, whereas the parasite assay is a 48 h proliferation assay that may depend on the extent to which the parasites can tolerate a purine deficit. Furthermore, the IC₅₀ value in a parasite strain will depend on the level of PfENT1 expression in that strain. Higher PfENT1 expression will result in a large IC₅₀ value in order to achieve the same level of inhibition of purine uptake in the proliferation assay.

Figure 6.

IC₅₀ curves for concentration-dependent inhibition of indicated P. falciparum parasite strain growth. WT is 3D7 strain, ART^R is artemisinin resistant strain (Cam3.II), HB3 contains the F36S/V129I SNP, and 7G8 contains the Q284E SNP. Results show the average of n≥3 biological replicate experiments. Mean ± SEM is shown. Average IC₅₀ values are in Table 3.

Figure 7.

Ratio of IC₅₀ values (n=3) for inhibition of P. falciparum strain proliferation by PfENT1 inhibitors to IC₅₀ values for inhibition of proliferation in wild-type P. falciparum (WT). Dashed line indicates a ratio of 1 and no difference in the IC₅₀ values. WT IC₅₀ values used are republished with permission from Frame et al., 2015.¹⁴ Symbols represent the values of individual experiments. Bars show mean ± SEM. Significantly different than WT by one-way ANOVA *(p ≤ 0.05), **(p ≤ 0.01), ***(p ≤ 0.001), ****(p ≤ 0.0001).

Table 3.

IC₅₀ values for concentration-dependent inhibition of parasite proliferation in indicated strains by PfENT1 inhibiting compounds. Average IC₅₀ values are from three independent trials.

IC50 values for concentration-dependent inhibition of proliferation in strains by compounds (μM)
Compound	WT	HB3 (F36S/V129I)	7G8 (Q284E)	ARTR
1	34 ± 2	23 ± 5	16 ± 0.2	29 ± 11
2	33 ± 8	17 ± 1*	9 ± 0.2**	12 ± 4**
3	21 ± 2	12 ± 2**	19 ± 0.8	7 ± 2***
4	14 ± 0.5	10 ± 2	3 ± 0.2**	8 ± 2
5	36 ± 0.2	31 ± 0.7	15 ± 1***	18 ± 5***
6	32 ± 0.9	23 ± 4	12 ± 0.8**	13 ± 4**
7	45 ± 2	24 ± 2**	12 ± 1***	19 ± 7**
13	7 ± 0.1	5 ± 0.9	4 ± 0.2*	3 ± 0.9**
19	45 ± 1	21 ± 5***	12 ± 2****	13 ± 3***

Data show mean ± SEM. WT IC₅₀ values are republished with permission from Frame et al., 2015.¹⁴ Significantly different than WT by one-way ANOVA,

^*(p ≤ 0.05),

^**(p ≤ 0.01),

^***(p ≤ 0.001),

^****(p ≤ 0.0001).

Curiously, there was also relatively little correlation between SNPs that altered purine affinity and SNPs that altered inhibitor potency in the yeast and parasite assays. For most of the SNPs that altered the affinity of one or more of the purines (L197F, A351S, F394L) there was no effect on inhibitor potency. This might suggest that the purines and inhibitors do not bind at the same site. However, we cannot exclude the possibility that these SNPs are not located at a binding site, but rather alter protein conformation to exert their effects of purine affinity. Only the Q284E mutation decreased the affinity for both the nucleobases and all of the inhibitors (Table 2, 3 and Figure 5). However, even the Q284E mutation did not significantly alter the affinity for the corresponding nucleosides. In the absence of a high resolution structure, it is difficult to explain these differences in the effects of the mutations.

Previously, we tested a subset of the 9 compounds (Compound 1, 2, 3, 5 and 13) against pfent1Δ parasites generated in Dd2 strain parasites (Frame et al., 2015). Compounds 1, 2, 3 and 13 had a 2- to 4-fold increase in IC₅₀ against pfent1Δ parasites compared to WT Dd2 parasites. Compound 5 had similar IC₅₀ values against both WT and pfent1Δ parasites. This suggests that PfENT1 inhibitors can kill parasites by hitting a secondary target. This might potentially mask any PfENT1 SNP effects, although for four of the five compounds tested the interaction with the secondary target was a lower affinity interaction. Frame et al., 2010 also observe a delayed-death phenotype in pfent1Δ parasites against compounds tested. pfent1Δ parasites had a progressive decrease in IC₅₀ values whereas WT parasites did not. This demonstrates that WT parasite killing is primarily due to inhibition of PfENT1 and not by inhibiting a secondary target. Currently, the secondary target is unknown. Our results demonstrate that parasite strains containing SNPs in PfENT1 are not resistant to the PfENT1 inhibitors tested here. These results show PfENT1 inhibitors are good starting points for further anti-malarial development.

Summary/Conclusion

PfENT1 is a promising target for antimalarial drug development. We have previously identified PfENT1 inhibitors that kill malaria parasites in culture. Here we have shown using our yeast expression system that naturally occurring non-synonymous SNPs in PfENT1 identified in field isolates have minimal effects on purine affinity and inhibitor potency. Furthermore, parasite strains containing a subset of the SNPs are not resistant to the PfENT1 inhibitors.

1. Four PfENT1 SNP mutations, L197F, Q284E, A351S, and F394L decreased the transporter affinity for one or more of the purines tested. Only the L197F and Q284E mutations altered the affinity for hypoxanthine, the major purine substrate for blood stage parasites.

2. The majority of the PfENT1 SNPs did not have an effect on PfENT1 inhibitor sensitivity in the yeast model.

3. The Q284E mutation reduced the apparent affinity of all of the PfENT1 inhibitors by 2- to 5-fold. The F36S mutant reduced the affinity of a subset of the chemotype B inhibitors and also chemotype E. However, parasite strains containing these mutations did not exhibit decreased sensitivity to the inhibitors.

4. PfENT1 inhibitor potency was similar or increased up to 10-fold in both drug sensitive and resistant parasite strains. This suggests that PfENT1 inhibitors are effective against parasites that are resistant to other antimalarial drugs.

5. PfENT1 inhibitors are good starting points for further drug development using hit-to-lead medicinal chemistry.

Materials and Methods

PfENT1 SNP DNA plasmid constructs and Yeast Transformation

Non-synonymous SNPs in the gene encoding PfENT1 (PF3D7_1347200) were identified in gene sequences from P. falciparum parasite strains contained in the PlasmoDB database (http://www.plasmodb.org/plasmo/). The mutants were generated using the QuickChange site directed mutagenesis kit (Agilent Technologies) in the yeast codon optimized pfent1 gene in the pCM189 yeast episomal expression plasmid described previously.¹⁴ Point mutations were verified using DNA sequencing. The plasmids were transformed into purine auxotrophic yeast by procedures described previously.¹⁴ Generation of the purine auxotrophic yeast was described previously by deletion of the ade2 gene (ade2Δ), which codes for a critical enzyme in the yeast’s de-novo purine biosynthesis pathway, in the BY4741 background (genotype MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0, fui1Δ::PfENT1).¹³

Yeast growth media

Yeast were grown in synthetic defined media (SDM) containing 0.17% (w/v) yeast nitrogen base (US Biologicals cat# Y2030), 0.02% (w/v) drop-out mix (US Biologicals cat# D9542; minus histidine, tryptophan, uracil, adenine, without nitrogen base), 2% (w/v) dextrose, 0.5% (w/v) ammonium sulfate, 40 mg/L histidine, 40 mg/L tryptophan and 10 mM adenosine.

Radiolabel substrate uptake inhibition by purines

To measure purine uptake competition, a 96-well plate (flat bottom 96 well clear, non-sterile polystyrene plate, Fisher Scientific, #12565501) were preloaded with 50 μL purines (hypoxanthine, adenine, inosine, or adenosine) serially diluted 3–fold in phosphate buffered solution (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM KH₂PO₄, 10 mM Na₂HPO₄, pH 7.4) for a final purine concentration range of 0.61μM to 6.35 mM. 50 μL of 1 μM [³H]uridine ([2,8- ³H]uridine; 22 Ci/mmol) (Moravek Biochemicals, Brea, CA) in PBS was added to each well and resuspended. Yeast were grown overnight in yeast growth media to mid-log phase, pelleted, and washed three times with PBS. Yeast cells were resuspended in PBS to a final concentration of 2 × 10⁸ cells/ml (cell density was determined as previously described in Frame et al., 2015). 100 μL yeast were added to each well and incubated at room temperature for 15 min. At the end of the time course, yeast were harvested onto glass fiber filter mats (Filtermat A, GF/C; Perkin Elmer) using a TomTec 96-well cell harvester system (#96-3-469). Filtermats were dried and sealed in plastic bags with 5 mL Betaplate Scint (Perkin Elmer) scintillation fluid. Counts were measured using a microplate scintillation counter (1450 Microbeta Trilux, Perkin Elmer).

Radiolabel uptake inhibition by compounds

Previously identified PfENT1 inhibiting compound were purchased from Chembridge (San Diego, CA).¹³ To maintain consistency with our previous publications, we refer to the compounds by the number of their rank order potency in the original HTS.¹⁴ The concentration dependence of their inhibition of [³H]adenosine uptake into PfENT1-expressing yeast was tested as described previously in Frame et al., 2015. Briefly, 96-well plates were preloaded with 100 μL of 100 nM [³H]adenosine ([2,8 – ³H]adenosine; 23 Ci/mmol; Moravek Biochemicals, Brea, CA) in PBS. 0.5 μL of 25 mM compound serially diluted 4-fold in DMSO was added to each well. 100 μL yeast with a concentration of 2 × 10⁸ cells/ml in PBS were added to each well and incubated for 15 min. At the end of each time course, cells were harvested and radioactivity counted as previously described.

Compound cytotoxicity assay with P. falciparum parasites

Concentration dependent cytotoxicity of compounds was tested in the 7G8, HB3, and artemisinin resistant Cam3.II parasite strains. Ring-stage synchronized cultures were used for all assays. Cultures were pelleted and resuspended in malaria culture media containing 10 μM hypoxanthine to 1% hematocrit and 2% parasitemia as described previously in Frame et al., 2015. 250 μL of culture was added to 96-well cell culture plates (Corning, cat# 3596) preloaded with 1.87 μL of 25 mM compound serially diluted 4-fold in DMSO. Plates were loaded into a sealed gas chamber and incubated at 37 °C for 48 h with 90% N₂/5% CO₂/5% O₂. After 48 h growth, parasite viability was assessed using SYBR Green I (Invitrogen, cat# S7567) DNA fluorimetry as previously described.¹⁴ The 7G8 (BEI MRA-154) and HB3 (BEI MRA-155) parasite strains were obtained from the Malaria Research and Reference Reagent Resource Center (MR4) managed by BEI Resources. The artemisinin-resistant Cam3.II strain was obtained from Dr. David Fidock (Columbia Univ.).

SNP genotyping in P. falciparum parasite lines HB3 and 7G8

Parasite lines HB3 and 7G8 were grown to a high parasitemia. Subsequently, we extracted genomic DNA using a Qiagen DNeasy blood and tissue kit. Genomic DNA samples were provided to Genewiz for SNP genotyping and mutation analysis. Genewiz Sanger sequencing results confirm the existence of SNP F36S/V129I in the HB3 parasite line and SNP Q284E in the 7G8 parasite line.

Computational and Statistical Analysis

IC₅₀ values were calculated from concentration-response data using GraphPad Prism log(inhibitor) vs. response - Variable slope (four parameters) function, Y=Y_max/(1+10^((LogIC₅₀-[X])*n_H)) where Y = effect or % of maximal signal, Y_max = maximal effect or 100% of maximal signal, IC₅₀ is the concentration causing half-maximal inhibition, [X] = concentration of compound X cause effect Y, n_H = Hill slope.

Statistical significance was determined using GraphPad Prism one-way analysis of variance (ANOVA).

Abbreviations.

ACT

Artemisinin-based Combination Therapy

ART^R

artemisinin-resistant P. falciparum strain

DHA

dihydroartemisinin

ENT

equilibrative nucleoside transporter

GMS

Greater Mekong subregion

HTS

high throughput screen

IC₅₀

concentration causing 50% inhibition of maximum

PBS

phosphate buffered solution

PbENT1

Plasmodium berghei ENT1

PfENT1

Plasmodium falciparum ENT1

PP

piperaquine

PvENT1

Plasmodium vivax ENT1

PfENT1-KO

PfENT1 knockout

SD

standard deviation

SEM

standard error of the mean

SDM

synthetic defined media

SNP(s)

single nucleotide polymorphism(s)

WT

wild type