Artemisinin resistance phenotypes and K13 inheritance in a Plasmodium falciparum cross and Aotus model

Significance

Artemisinin-based combination therapies (ACTs) are first-line antimalarial therapies used worldwide. The artemisinin drug (ART) component clears the bulk of infection rapidly, but small numbers of persistent parasites must be removed by the partner drug. Longer parasite clearance t_1/2 values have been associated with a Kelch-propeller mutation (K13 C580Y), raising concerns of increased ART resistance. We investigated effects of C580Y by using a Plasmodium falciparum cross and a monkey malaria model. Following three standard doses of ART, infections with or without the C580Y mutation cleared to microscopically undetectable levels, as in humans; however, frequent recrudescences occurred with both types of infection. These results emphasize the importance of effective partner drugs to kill the parasites that persist through the ART component of ACT.

Abstract

Concerns about malaria parasite resistance to treatment with artemisinin drugs (ARTs) have grown with findings of prolonged parasite clearance t_1/2s (>5 h) and their association with mutations in Plasmodium falciparum Kelch-propeller protein K13. Here, we describe a P. falciparum laboratory cross of K13 C580Y mutant with C580 wild-type parasites to investigate ART response phenotypes in vitro and in vivo. After genotyping >400 isolated progeny, we evaluated 20 recombinants in vitro: IC₅₀ measurements of dihydroartemisinin were at similar low nanomolar levels for C580Y- and C580-type progeny (mean ratio, 1.00; 95% CI, 0.62–1.61), whereas, in a ring-stage survival assay, the C580Y-type progeny had 19.6-fold (95% CI, 9.76–39.2) higher average counts. In splenectomized Aotus monkeys treated with three daily doses of i.v. artesunate, t_1/2 calculations by three different methods yielded mean differences of 0.01 h (95% CI, −3.66 to 3.67), 0.80 h (95% CI, −0.92 to 2.53), and 2.07 h (95% CI, 0.77–3.36) between C580Y and C580 infections. Incidences of recrudescence were 57% in C580Y (4 of 7) versus 70% in C580 (7 of 10) infections (−13% difference; 95% CI, −58% to 35%). Allelic substitution of C580 in a C580Y-containing progeny clone (76H10) yielded a transformant (76H10^C580Rev) that, in an infected monkey, recrudesced regularly 13 times over 500 d. Frequent recrudescences of ART-treated P. falciparum infections occur with or without K13 mutations and emphasize the need for improved partner drugs to effectively eliminate the parasites that persist through the ART component of combination therapy.

Artemisinin-based combination therapies (ACTs) are first-line antimalarial treatments comprising a short-acting artemisinin drug (ART) for rapid parasitemia reduction plus a long-acting partner drug to eliminate surviving parasites that can cause recrudescences. The need for effective ART partners was recognized in early clinical trials showing recrudescence within 28 d in more than 40% of Plasmodium falciparum-infected patients treated with 3-d artemisinin (also known as qinghaosu or QHS) monotherapy (1). These recrudescences reflected properties of the parasites before large-scale use of ART and were classified by Li et al. (1) as RI-resistant by World Health Organization (WHO) criteria (microscopic clearance followed by recrudescence within 28 d) (2). Frequent recrudescences were subsequently reported from other studies with 3–7 d of ART monotherapy (3, 4), including 16 of 111 P. falciparum-infected patients treated in Vietnam with QHS doses of 10 mg/kg on day 1 followed by 5 mg/kg/d on days 2–7 (5).

ACTs with appropriate partner drugs should be >90% effective (6), but successful treatments are threatened by the continual evolution of drug-resistant Plasmodium. For example, dihydroartemisinin (DHA; the active metabolite of QHS) plus piperaquine (PPQ), an ACT with a generally good track record, now fails frequently in regions of Cambodia, leading to the replacement of DHA/PPQ by artesunate (AS) plus mefloquine (MEF) (6–11). Partner drug ineffectiveness allows selection pressure for ART resistance, as do persistent use of monotherapy and poor-quality medicines (12). In the Greater Mekong Subregion (GMS), these concerns have grown with reports of prolonged P. falciparum parasite clearance time (PCT; i.e., the time required for a patient to show negative blood smear results) after AS monotherapy or ACT. ART typically clears ring-stage parasites within 48 h (4), but longer clearances often occur (1, 3), and PCTs of 72–96 h may predominate in regions of the GMS (13–15). Clearance t_1/2 (i.e., the expected time to reduce 50% of parasitemia) generally ranges from less than 3 h to more than 7 h (16).

In the 1990s, the RI, RII, and RIII descriptors of resistance (2) were updated by WHO to new definitions of early treatment failure within 3 d and late clinical or parasitological failure within 14–28 d (17, 18). Increasing levels of resistance to ART itself have not been demonstrated based on any of these definitions. Such resistance is instead suspected when patients have certain mutations of P. falciparum protein K13 (encoded by gene pfk13) and a clearance t_1/2 >5 h or microscopically observed parasitemia 3 d after start of treatment (6). Whether these are indicators of increasing clinical resistance to ART and portend its failure remains controversial (19–26). In studies with ART monotherapy, prolonged PCTs or day 3 parasite-positive rates did not associate with recrudescences within 42 d (27, 28). PCTs (or clearance t_1/2 values) likewise could not be associated with delayed resolutions of malaria fever after monotherapy (14, 27, 29).

PCT and t_1/2 variations in ART-treated infections do not correlate with ART IC₅₀ values of corresponding P. falciparum isolates tested for 48–72-h periods in vitro (14, 27, 30). However, associations occur between t_1/2 values and in vitro survival levels of early ring-stage parasites (0–3 h postinvasion) exposed to a pulse of physiological-dose DHA in ring-stage survival assays (RSAs) (31). Differential ring-stage survival and its association with t_1/2 led to proposals that doubled-frequency ART dosing might improve treatment outcome, but this was not observed in a clinical trial (32), likely because of the high sensitivity of trophozoites that develop from ring stages during the overall course of treatment (31, 33).

Polymorphisms in the Kelch propeller domain of P. falciparum protein K13 are associated with ACT failures (8–10) and with longer vs. shorter t_1/2 values in the GMS (10, 16, 34). K13 polymorphisms such as C580Y, Y439H, M476I, R539T, and I543T also increase ring-stage survival in RSAs (35, 36). Studies of ACT failures, however, can be confounded by the difficulties of distinguishing partner drug from ART resistance, as well as low levels of recombination among P. falciparum genotypes (37). For example, strong associations of K13 C580Y and Y493H are reported for in vitro measures of PPQ resistance (defined by >10% rates in PPQ survival assays) (38, 39); these associations in phenotypes that involve multiple genes, e.g., pfcrt, pfmdr1, and pfpm2 (40), raise further questions about the function of K13 and roles it may have in various developments of ACT resistance.

Evaluation of ART responses in a controlled in vivo setting can help to answer questions about clearance t_1/2 values, recrudescences, ring-stage survival effects, and K13 polymorphisms, but this requires the availability of suitable animal models or controlled human infections. Aside from the Great Ape Pan troglodytes (i.e., chimpanzee), animal models for human malaria have long included the New World monkeys Aotus and Saimiri (41–43). Here we describe the generation of a P. falciparum cross for progeny infective to Aotus monkeys and report analyses of the in vitro and in vivo phenotypes involved in ART resistance.

Results

Genetic Cross to Investigate P. falciparum Artemisinin Resistance in Vivo.

Various P. falciparum lines have been used in Aotus nancymaae for drug and vaccine development (42), but no K13 C580Y parasites have been adapted to these monkeys to our awareness (Dataset S1). To obtain C580Y parasites capable of infecting Aotus and take advantage of this model, we crossed the Aotus-noninfective 803 parasite line (30), isolated from a C580Y Cambodian infection with an 11-h t_1/2 during AS treatment of 4 mg/kg/d [by the WorldWide Antimalarial Resistance Network’s (WWARN’s) clearance estimator (44)], with parasite GB4, an Aotus-infective C580 clone from a Ghanaian isolate obtained before widespread use of ART in Africa (45) (Fig. 1A). DHA IC₅₀ values for 803 and GB4 were 1.83 ± 0.09 nM and 3.16 ± 0.12 nM (mean ± SEM; Fig. 1B and SI Appendix, Table S1), consistent with reports of no increase in IC₅₀ values with the C580Y change (34). RSA survival for 803 (9.52 ± 1.32%) was 13-fold higher than for GB4 (0.72 ± 0.40%; Fig. 1C and SI Appendix, Table S1), similar to those of other parasites with this K13 difference (34–36).

Fig. 1.

Design of the P. falciparum 803×GB4 cross and in vitro drug responses. (A) Mixed gametocytes of the parents were fed to Anopheles mosquitoes, and resulting sporozoites were inoculated into three Sp− NHPs: A. nancymaae (night or owl monkey), P. troglodytes (chimpanzee), and S. boliviensis (squirrel monkey). Blood-stage recombinant progeny were recovered only from the chimpanzee; after cultivation in human erythrocytes, some progeny were able to infect Aotus. (B) DHA IC₅₀ values from 803 and GB4 parents (gray), 20 progeny (black) including two pairs of isogenic clones (curly braces), a GB4 self-mating product 34F5, and the engineered 76H10^C580Rev revertant (checkerboard pattern). K13 types are indicated. (C) RSA results from 803×GB4 parasites and the engineered 76H10^C580Rev revertant. Details of the data are listed in SI Appendix, Table S1.

To generate the 803×GB4 cross, gametocytes were induced in vitro, mixed, and fed to Anopheles mosquitoes (Fig. 1A). Cross-fertilization was verified by analysis of single oocysts dissected from mosquito midguts (SI Appendix, Fig. S1A). Sporozoites were recovered from mosquito salivary glands, cryopreserved, and confirmed to carry markers of both parents (SI Appendix, Fig. S1B) before inoculation into a splenectomized (Sp−) P. troglodytes chimpanzee and two Sp− New World monkeys (A. nancymaae and Saimiri boliviensis). Recombinant haploid blood-stage progeny developed from sporozoites only in the chimpanzee and were collected 18–40 d after inoculation (SI Appendix, Figs. S1C and S2A). No progeny parasite was recovered from sporozoite inoculations of Aotus or Saimiri monkeys, suggesting limited parasite development in the liver in consequence of the evolutionary distance between these New World monkeys and the Great Apes. However, consistent with the findings of a previous study (45), parasites from the 803×GB4 cross that had inherited the GB4-type PfRH5 erythrocyte invasion ligand were able to infect Aotus by direct bloodstream inoculation (Dataset S1). Limiting dilution cloning yielded more than 400 progeny (SI Appendix, Fig. S2B). Microsatellite fingerprinting was used to preliminarily sort these progeny by marker inheritance patterns into segregant types; further analysis by DNA microarrays assessed 3,629 SNPs and identified groups represented by a self-mating product of GB4 (34F5) and 27 independent recombinants (SI Appendix, Fig. S3 and Dataset S2).

Artemisinin Ring-Stage Survival, but Not IC₅₀, Linked to K13 C580Y Inheritance.

DHA IC₅₀ values of 20 arbitrarily chosen progeny ranged from 0.9 to 4.4 nM (Fig. 1B); geometric means (GMs) of these IC₅₀ values were similar for the C580Y clones and C580 clones [ratio of GM(C580Y)/GM(C580) = 1.00; 95% CI, 0.62–1.61]. By quantitative trait loci (QTL) analysis (SI Appendix, Fig. S4), no significant association was found for these IC₅₀s with any genome region. In contrast, RSA results clustered into distinct C580Y and C580 groups [ratio of GM(C580Y)/GM(C580) = 19.56; 95% CI, 9.76–39.21], and this RSA phenotype mapped to a 300-kb region containing the pfk13 gene in chromosome 13 (Fig. 1C and SI Appendix, Fig. S5 and Table S2). The role of K13 in the RSA phenotype was confirmed by allelic modification of 76H10 into a transformant clone, 76H10^C580Rev, carrying a C580Y→C580 change accompanied by a phenotype switch of ring survival from >10% to <2% (Fig. 1C). Heritable modulators of the main K13 determinant likely contribute to variations in the high- or low-RSA survival groups; however, linkage analysis of >200 independent recombinant progeny may be needed to capture modulating loci, as has been done in some cases for yeast (46).

Recrudescences With or Without the K13 C580Y Mutation in AS-Treated Aotus.

We next evaluated AS responses of 17 primary infections in Sp− Aotus monkeys with 803×GB4 parasites carrying K13 C580Y or C580 (Fig. 2 A–C, Table 1, and Dataset S1). Seven of these infections were C580Y-type (87E7, 76H10, and 85G7) and 10 were C580-type (GB4, 61D3, and 76H10^C580Rev). Treatment with three i.v. injections of AS 4 mg/kg/d (15 infections) or 2 mg/kg/d (2 infections) promptly reduced the parasitemias of all infections to subpatent (i.e., microscopically undetectable) levels, as observed in human studies (28). Three of the seven C580Y-type infections recrudesced within 28 d and another recrudesced at 35 d posttreatment, but all of these spontaneously cleared to subpatent levels (CS; Table 1). Of the 10 C580-type infections, 7 recrudesced within 28 d: 4 required repeat treatments and 3 spontaneously cleared (Table 1). Microscopically observed recrudescence rates were therefore 57% (4 of 7, all 4 CS) in the C580Y infections vs. 70% (7 of 10, 3 CS) in the C580 infections (difference, −13%; 95% CI, −58% to 35%). Remarkably, one C580 infection, 76H10^C580Rev/86236, recrudesced regularly over 500 d, providing clearance information from 14 AS treatments (Fig. 2C); recrudescences 4, 8, 9, and 10 of this series were checked and showed no acquisition of a K13 mutation (SI Appendix, Fig. S6). Although the recrudescence rate difference estimate is not precise (Dataset S3, power calculations), the stronger recrudescences observed with C580 infections hint that C580Y could carry a fitness cost in the P. falciparum genome.

Fig. 2.

AS responses of 803×GB4 parasites in Aotus. Information in these panels corresponds to entries in Table 1 except for the data from FVO infections. (A–C) Primary and recrudescent parasitemias in Aotus monkeys receiving AS treatment: 87E7, 76H10, and 85G7 infections (K13 C580Y-type); GB4 and 61D3 infections (K13 C580-type); and 76H10^C580Rev revertant infections (K13 C580-type). Numbers after parasite names identify individual Aotus. AS1, AS treatments of primary infections; AS2–AS14, treatments of recrudescences after AS1; CS, cleared to subpatent levels without treatment. “T2” identifies treatment of a 61D3 recrudescence with a different antimalarial regimen; “MEF” indicates administration for cure as a single 25-mg/kg oral dose. (D) Ring-stage parasite clearance curves from AS-treated Aotus infections presented as percentage of initial count. C580Y-type primary clearance profiles are shown in red, and C580-type primary and recrudescence clearances are in gray and blue, respectively. Few trophozoite and schizont stages were observed in the circulation, and clearance results were similar when these were included with the ring-stage counts for analysis (total parasitemia; Dataset S3). (E) Plot of ring-stage parasite clearance t_1/2 values obtained by the WWARN (“W”), joinpoint (“J”), and AHL95 (“A”) methods for 803×GB4 parasites carrying the C580Y (red) or C580 polymorphism (gray, primary clearances; blue, recrudescence clearances). Dotted line represents the 5-h t_1/2 threshold proposed by WHO (6).

Table 1.

P. falciparum infections and AS treatment outcomes in Aotus

Parasite line	K13 580 residue	Aotus ID	Spleen status	Previous infection?	Prepatent period, d	AS dose, mg/kg/d × 3	Initial P, %	Recrudescence information	WWARN t1/2 (ring-stage)	Joinpoint t1/2 (ring-stage)	AHL95 t1/2 (ring-stage)
GB4	C	86329	Sp−	Yes, FVO	7	4	1.63	NO	4.61	5.85	4.91
GB4	C	86355	Sp−	No	10	4	0.74	25 DPT; CS	ND	7.31	7.20
GB4	C	86489	Sp−	No (PINP)	10	4	1.37	11 DPT	17.36	7.03	7.85
						4	2.45	ND; “R1”	9.27	7.20	5.76
GB4	C	86574	Sp−	Yes, FVO	7	4	1.13	NO	6.79	4.57	4.39
61D3	C	86354	Sp−	No	17	4	2.68	NO	13.00	10.48	6.93
61D3	C	86484	Sp−	No (PINP)	3	2	3.78	11 DPT	5.95	8.45	6.58
87E7	Y	86349	Sp−	No	15	4	3.05	NO	8.07	7.72	6.90
85G7	Y	AI2045	Sp−	Yes, Pv	10	4	1.14	18 DPT; CS	7.76	9.96	8.23
85G7	Y	AI2049	Sp−	Yes, Pv	10	4	0.96	19 DPT; CS	7.20	9.73	10.04
76H10	Y	86416	Sp−	Yes, FVO	5	4	1.31	NO	4.71	8.46	6.58
76H10	Y	86458	Sp−	No (PINP)	19	4	1.14	NO	5.91	9.66	7.40
76H10	Y	86527	Sp−	No	6	4	1.13	20 DPT; CS	11.10	8.69	8.39
76H10	Y	86564	Sp−	No	6	2	0.96	35 DPT; CS	10.98	7.75	8.12
76H10C580Rev	C	86236	Sp−	Yes, Pv	24	4	2.03	23 DPT	6.32	10.78	4.54
						4	3.80	15 DPT; “R1”	5.34	9.45	6.72
						4	1.49	19 DPT; “R2”	10.60	8.47	6.73
						4	2.41	17 DPT; “R3”	8.04	7.43	7.54
						4	1.26	18 DPT; “R4”	ND	7.08	7.30
						4	2.59	21 DPT; “R5”	4.71	7.44	7.17
						4	1.36	23 DPT; “R6”	9.41	10.40	8.49
						4	1.22	29 DPT; “R7”	11.69	5.57	7.17
						4	1.84	26 DPT; “R8”	6.83	7.56	6.23
						4	1.03	40 DPT; “R9”	8.84	8.31	8.65
						4	0.96	37 DPT; “R10”	11.89	8.05	7.46
						4	2.05	56 DPT; “R11”	6.70	7.92	5.76
						4	1.03	33 DPT; “R12”	10.42	10.87	12.82
						4	0.90	NO; “R13”	6.16	9.70	8.81
76H10C580Rev	C	86381	Sp−	Yes, FVO	6	4	2.05	22 DPT; CS	4.90	7.19	6.05
76H10C580Rev	C	86383	Sp−	Yes, Pv	26	4	1.26	25 DPT	6.20	7.31	4.10
						4	0.87	31 DPT; “R1”	9.27	9.22	7.25
76H10C580Rev	C	86404	Sp−	No (PINP)	42	4	1.45	18 DPT; CS	6.45	11.54	6.32
FVO	C	86329	Sp+	No	2	4	0.95	17 DPT	7.95	4.16	4.95
FVO	C	86381	Sp−	No	2	4	0.99	10 DPT	6.52	8.33	7.28
FVO	C	86416	Sp−	No	2	4	1.11	7 DPT	12.97	6.55	6.89
FVO	C	86436	Sp+	No	4	4	8.97	10 DPT	5.17	6.86	5.89
FVO	C	86574	Sp+	No	2	4	0.81	7 DPT	5.99	3.87	5.29

CS, cleared to subpatent levels without antimalarial treatment; DPT, days posttreatment; NO, not observed by microscopic evaluation; ND, not determined; PINP, previously inoculated with Plasmodium but did not develop parasitemia; P, parasitemia; Pv, Plasmodium vivax; “R1”–“R13”, labeled recrudescences after treatments AS1–AS13 (Fig. 2 B and C).

Little or No Effect of K13 C580Y on Parasite Clearance t_1/2 in AS-Treated Aotus.

Parasite clearance curves of the 17 primary and 15 recrudescent 803×GB4 infections in AS-treated Aotus presented comparable traces (Fig. 2D). Notably, these traces do not show the wide separation of slow or fast clearance that have been associated with the K13 polymorphisms in human populations of the GMS (34). By the WWARN parasite clearance estimator (44), the average ring-stage t_1/2 values were 7.96 h for the 7 primary C580Y infections and 7.95 h for 9 of the 10 primary C580 infections (0.01-h difference; 95% CI, −3.66 to 3.67; the WWARN algorithm was unable to analyze data from 1 clearance). We also developed two alternative methods to calculate t_1/2 values: (i) a joinpoint method like WWARN’s but designed to accommodate readings by multiple microscopists and (ii) a method based on the time to 95% reduction of initial parasitemia (AHL95; Dataset S3). By joinpoint calculation, ring-stage t_1/2 estimates were 8.85 h for the 7 primary C580Y infections and 8.05 h for the 10 primary C580 infections (0.80-h difference; 95% CI, −0.92 to 2.53). By the AHL95 method, a small but significant separation was found between 7.95 h for the 7 primary C580Y infections and 5.89 h for the 10 primary C580 infections (2.07-h difference; 95% CI, 0.77–3.36). Despite the differences in these estimates, we note that all three methods yielded t_1/2 values predominantly longer than 5 h (Fig. 2E) and that they captured only the rate of bulk clearance, ignoring the tails of low parasitemia, which are potentially important for recrudescence.

As antimalarial immunity can affect observed t_1/2 (47), we analyzed the C580Y and C580 clearances for possible effects from previous Aotus exposure to malaria parasite infection: adjustment for previous infection by a linear model made no substantial modifications to our findings (Dataset S3). The results from Sp− animals led us to compare AS-treated infections in spleen-intact (Sp+) and Sp− Aotus with P. falciparum FVO (Vietnam Oak Knoll), a C580-type parasite line that can infect Sp+ animals and is widely used for vaccine and drug studies (42, 43). In treated FVO infections of three Sp+ and two Sp− monkeys (SI Appendix, Fig. S7), the difference of mean t_1/2 values by the WWARN estimator was −3.38 h (95% CI, −35.59 to 28.84; Dataset S3). Tighter CIs were obtained by the joinpoint and AHL95 methods, i.e.,−2.47 h (95% CI, −6.82 to 1.87) and −1.71 h (95% CI, −2.79 to −0.62), respectively (Table 1 and Dataset S3). Recrudescences occurred in all FVO-infected Aotus, whether Sp+ or Sp− (Table 1).

Frequent recrudescences of C580 infections in AS-treated Aotus are consistent with the intrinsic R1 resistance of P. falciparum originally reported from ART monotherapy studies in humans (1). Dormancy is hypothesized as a mechanism for the survival of small numbers of parasites, which may enter a quiescent state and resume growth later with no change of phenotype (48–50). In agreement with this hypothesis, our data showed no changes in the RSA survival or IC₅₀ values of parasites from Aotus infections before and after AS treatment (SI Appendix, Fig. S8 A–C). In assays of in vitro recrudescence by a DHA plus sorbitol method (51), the average time to return to 2% parasitemia was 28.5 d for 76H10 C580Y parasites (95% CI, 25.19–31.81 d) vs. 23.0 d for 76H10^C580Rev parasites (95% CI, 18.03–27.97 d; difference of 4.67 d; P = 0.02; SI Appendix, Fig. S8D).

Discussion

Our results show that the P. falciparum–Aotus model recapitulates many features of ART responses in humans, including similar effectiveness of 2- or 4-mg/kg/d AS treatments, fast reductions to subpatent parasitemia, and high rates of recrudescence in vivo within 28 d. As in humans, these recrudescences occurred with K13 C580- as well as C580Y-type infections, and parasites from these recrudescences showed no change of K13 sequence or in vitro phenotypes, even after multiple rounds of ART treatment.

RSA differences are readily observed between synchronized C580Y and C580 ring-stage parasites exposed to a 6-h pulse of 700 nM DHA (physiological concentration). Whereas 10–20% of C580Y rings survive this DHA pulse, only 0.1–2.0% of C580 rings do so. Later trophozoite and schizont stages exhibit no K13-mediated differences of survival in response to the DHA pulse, which is approximately 1% regardless of K13 type (31). These similar trophozoite and schizont susceptibilities account for the efficacies of 3-d AS monotherapy against C580Y and C580 infections and for the absence of a signal from our QTL search of DHA IC₅₀ values (obtained from parasites exposed to the drug for 72 h).

The absence of a distinct separation between the clearance t_1/2 values of AS-treated K13 C580Y and C580 infections in Aotus may reflect different host–parasite relationships in Sp− monkeys than in spleen-intact humans. We note that a splenic effect may be present in the clearance t_1/2 values of FVO infections in Sp+ vs. Sp− Aotus. Could it be that the spleen more effectively removes ART-treated C580 ring-stage forms than C580Y ring-stage forms, and that this differential removal capability is missing from Sp− Aotus? Such conjecture is consistent with a previous report that ART-damaged, unviable ring stages can retain their morphological appearance and circulate for several hours when there is no splenic clearance (52). Splenic recognition of ART-damaged parasites could thus account for shorter t_1/2 values of C580 ring-stage forms in spleen-intact humans vs. Sp− Aotus. Other aspects of immunity that influence P. falciparum clearance t_1/2 in humans (47, 53) could also differ in Aotus. As mentioned here earlier, a fraction of parasites that enter dormancy and escape clearance may account for the frequent recrudescences of C580-type and C580Y-type infections in Sp+ as well as Sp− animals.

The sample size of Aotus infections in our clearance t_1/2 determinations is small in comparison with the number of human t_1/2 values reported from surveys in the GMS. However, because the t_1/2 values reported from human C580- and C580Y-type infections were so different (e.g., figure 4A of ref. 34), we would have had greater than 99% power to detect a similar effect in our Aotus sample size (Dataset S3, power calculations). Although this study does not exclude a role for K13 in ART resistance, our findings are consistent with exceptions to the association of K13 C580Y with RSA and clearance t_1/2 (54–56), challenge the utility of K13 markers to predict treatment recrudescence, and raise important questions about interpretation of ring-stage clearance t_1/2 after ART treatment.

Considering the data from genome-wide association studies, we note that regional distributions of K13 SNPs in the GMS may include effects of ART selection not involving clinical resistance per se. For example, findings of greater gametocytogenesis with some K13 mutations (57) raise the intriguing possibility that P. falciparum transmission to mosquitoes may be protected as a survival mechanism not affecting clinical outcome (gametocytes, in contrast to asexual trophozoites and schizonts, cause no symptoms or recrudescences). K13 polymorphisms in P. falciparum populations may also arise from possible involvement in partner drug resistance and fitness adaptations accompanying the acquisition of new phenotypes.

Our results concur with reports that exceptions occur in regard to K13 associations with clearance t_1/2 values (54–56) and that R1 recrudescence rates after ART monotherapy do not correlate with PCTs (14, 27, 28, 32). Antimalarial experiences in the GMS recognize RI resistance to ART as a steady feature of the P. falciparum species, underscoring the importance of effective partner drugs to prevent ACT clinical failures. The recent recommendation of an ACT switch to AS/MEF where there is DHA/PPQ-resistant malaria in Cambodia (6, 11) is in line with these experiences. Preservation of ART as a linchpin of antimalarial therapy demands continued discovery and development of partner drugs that can ensure complete cure with no recrudescence of infection.

Methods

Generation of 803×GB4 Recombinant Sporozoites.

Cultures containing at least 0.5% stage V mature gametocytes were centrifuged at 2,500 rpm for 3 min in a Sorvall Legend RT centrifuge (Highconic Rotor 6 × 50 mL, no. 75003046) with settings of 37 °C, acceleration 5, and deacceleration 5 (Thermo Scientific). Pellets were suspended in 100% O+ human serum at 37 °C to 50% hematocrit. Mosquitoes were fed different mixtures of the 803 and GB4 gametocyte preparations (1:1, 2:1, 3:1, and 5:1; 400–600 µL total volume per mixture) by apposition of the netted mosquito containers to Parafilm membrane-covered glass feeders (Chemglass Life Sciences) maintained at 37 °C by a warm-water pump (Thermo Scientific). Each feeding was performed in a secure insectary with 50–100 female Anopheles stephensi Nijmegen mosquitoes (age 3–5 d) that had been starved of supporting 10% vol/vol corn syrup solution for 12–16 h. After subsequent maintenance of the mosquitoes with the corn syrup solution for 6–9 d at 26 °C and 75% humidity, 10–15 individuals from each feeding were randomly selected for midgut dissections, oocyst counts, and DNA extraction for PCR analysis. For all dissections, mosquitoes were transferred to a secure container, exposed to chloroform vapor (1 min), placed in 70% ethanol (1 min), and washed briefly in PBS solution (1.05 mM KH₂PO₄, 5.6 mM Na₂HPO₄, 154 mM NaCl, 2.7 mM KCl, pH 7.4; KD Medical). Midguts were removed with no. 5 tweezers and observed on a stereo microscope (Leica). For confirmation and counts, oocysts were stained with 0.05% mercurochrome solution in water (10 min) and visualized by using a bright-field microscope (10–40× objective). Single oocysts were removed from the midgut and transferred with a 23-gauge needle to 50 µL of homogenization buffer (80 mM EDTA, 100 mM Tris, pH 8.0, 160 mM sucrose) and frozen at −20 °C until DNA extraction and PCR analysis. After confirmation of oocysts from each feeding, the remaining mosquitoes were maintained until 15 d postfeeding, when they were transferred to Sanaria (Rockville, MD) for salivary gland dissections and sporozoite collection. From the four feedings, 370,000 sporozoites were purified and cryopreserved by using the Sanaria protocol (58); 50,000 of these were taken for genetic analysis (SI Appendix, Fig.S1), and 320,000 were used for inoculation of the chimpanzee. Parallel mosquito feedings with mixtures of 803 and GB4 gametocytes were also performed to obtain two additional batches of recombinant sporozoites for inoculation of New World monkeys: one containing 29,000 sporozoites was cryopreserved and used for an Sp− S. boliviensis monkey, and another containing 147,000 sporozoites was cryopreserved and used for an Sp− A. nancymaae monkey.

Animal Care, P. falciparum Blood-Stage Infections, and Treatments.

All nonhuman primate (NHP) care and use in this study was in accordance with the National Institutes of Health Animal Research Advisory Committee Guidelines, under protocols approved by the National Institute of Allergy and Infectious Diseases Animal Care and Use Committee, in compliance with the Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals (59). For clearance studies, monkeys with ≥0.5% parasitemia (mean, 1.88%; 95% CI, 1.34–2.41; Table 1) and no signs of self-clearance (stable and increasing parasitemia for ≥3–5 d) were treated with three daily i.v. AS injections. Thin and thick blood smears were prepared at the time of the first dose and then 3, 6, 12, 24, 30, 36, 48, 54, 72, and 96 h afterward. Detailed animal procedures are described in SI Appendix, Supplemental Methods Information, and a list of all NHP inoculations performed in this study is provided in Dataset S1.

Sporozoite Inoculations of NHPs and Blood-Stage Progeny Recovery.

Cryopreserved sporozoites were thawed and formulated for i.v. delivery as in trials of attenuated sporozoite vaccines in humans (58). Inoculation of the two batches of sporozoites into the Saimiri or Aotus monkeys produced no blood-stage parasitemias; however, recombinant progeny were obtained by inoculation of a single Sp− chimpanzee. Fourteen days after inoculation of 320,000 thawed sporozoites, blood-stage parasites were detected in the chimpanzee. Parasitemia increased to 0.04% on the 21st day postinoculation (DPI) and then decreased to a level detected only by PCR on the 25th DPI. Progeny pools (PPs) were collected in blood samples from the chimpanzee 18, 21, 22, 25, 28, 36, and 40 DPI (PP1–PP7; SI Appendix, Fig. S2). Portions of these pools were cryopreserved and adapted to cultures in human erythrocytes for cloning and Aotus inoculation. The chimpanzee was treated with a 500-mg dose of MEF 40 DPI to eliminate any remaining parasites and was released from the study.

Detailed Methods.

Parasite cultivation, progeny cloning and genotyping, pfk13 modification, IC₅₀ measurements, ring-stage survival, and in vitro recrudescence assays are described in detail in SI Appendix, Supplemental Methods Information. Further details of clearance t_1/2 determinations and statistical analysis are provided in Dataset S3.