Plasmodium falciparum rosetting protects schizonts against artemisinin

Abstract

Background: Artemisinin (ART) resistance in Plasmodium falciparum is thought to occur during the early stage of the parasite's erythrocytic cycle. Here, we identify a novel factor associated with the late stage parasite development that contributes to ART resistance.

Methods: Rosetting rates of clinical isolates pre- and post- brief (one hour) exposure to artesunate (AS, an ART derivative) were evaluated. The effects of AS-mediated rosetting on the post-AS-exposed parasite's replication and survival, as well as the extent of protection by AS-mediated rosetting on different parasite stages were investigated. The rosetting ligands, mechanisms, and gene mutations involved were studied.

Findings: Brief AS exposure stimulated rosetting, with AS-resistant isolates forming more rosettes in a more rapid manner. AS-mediated rosetting enabled infected erythrocytes (IRBC) to withstand AS exposure for several hours and protected the IRBC from phagocytosis. When their rosetting ability was blocked experimentally, the post-AS exposure survival advantage by the AS-resistant parasites was abrogated. Deletions in two genes coding for PfEMP1 exon 2 (PF3D7_0200300 and PF3D7_0223300) were found to be associated with AS-mediated rosetting, and these mutations were significantly selected through time in the parasite population under study, along with the K13 mutations, a molecular marker of ART-resistance.

Interpretation: Rapid ART parasite clearance is driven by the direct oxidative damages on IRBC by ART and the phagocytic destruction of the damaged IRBC. Rosetting serves as a rapid 'buying time' strategy that allows more parasites to complete schizont maturation, reinvasion and subsequent development into the intrinsically less ART-susceptible ring stage.

Funding: A*STAR, NMRC-OF-YIRG, HRC e-ASIA, Wellcome.

Keywords: Artemisinin resistance; PfEMP1; Plasmodium falciparum; rosetting.

Fig. 1

Brief AS-P. falciparum exposure. (a) The degree of increment in rosette (inset) formation post-AS exposure. AS-mediated rosette-stimulation range 12 – 500%. With the median of AS-mediated rosette-stimulation as cut-off, isolates formed two clusters (mean and S.D. shown) with different AS-mediated rosette-stimulation (Mann-Whitney P < 0·0001). (b) Long AS-PCt_1/2 had higher AS-mediated rosette-stimulation (range 60 – 420%; mean 200·3%, S.D. 100·8) than short AS-PCt_1/2 (range 14·29 – 50%; mean 32·272%, S.D. 12·73) (Mann-Whitney P < 0·0001). (c) Difference in the speed of AS-mediated rosetting between long- and short AS-PCt_1/2 groups to attain at least 50% of rosetting rate increment after AS exposure. The rosetting rates of isolates were monitored at 10-minutes interval until the 60^th minute post-AS exposure (indicated by dotted line in the plot). As none of the isolates with short AS-PCt_1/2 demonstrated 50% rosetting rate increment within the 60 minutes of monitoring (and demonstrated plateau trend on AS-mediated rosette-stimulation within this period, as shown by Supplementary Figure 3e), an arbitrary value of “70 minutes” were used for statistical comparison. On average, isolates with long AS-PCt_1/2 required 12·86 ± 4·688 minutes to reach 50% increment in their rosetting rates. Significant difference was found between the two groups (Mann-Whitney P < 0·0001). (d) Growth of parasites (at H₇₂) after 1 hr-AS exposure at different stages (ring, trophozoite and schizont) and trypsinised schizont (to prevent rosette formation by removing rosetting ligands on the surface of IRBC). Parasites with long AS-PCt_1/2 experienced significantly higher growth than the short AS-PCt_1/2 group (Welch's t test P = 0·0118, t = 2·825, df = 16·70; P < 0·0001, t = 5·868, df = 14·35; P = 0·0002, t = 4·964, df = 13·17 for exposure at ring, trophozoite and schizont stages, respectively). For the trypsinised schizont setting, no significant difference was found between the parasite growth of both AS-PCt1/2 groups (P = 0·3123, t = 1·051, df = 13). (e) Growth of parasites (at H₇₂) after 6 hrs-AS exposure at different stages (ring, trophozoite and schizont) and trypsinised schizont. Parasites with long AS-PCt_1/2 experienced significantly higher growth than the short AS-PCt_1/2 group (Welch's t test P = 0·0003, t = 4·579, df = 16·81; P = 0·0002, t = 4·938, df = 15·57; P = 0·0001, t = 5·196, df = 14·87 for exposure at ring, trophozoite and schizont stages, respectively). For the trypsinised schizont setting, no significant difference was found between the parasite growth of both AS-PCt1/2 groups (P = 0·5035, t = 0·691, df = 11·28). Sample size (n) is stated in each plot.

Fig. 2

AS-mediated rosetting and phagocytosis. (a) AS did not alter THP-1’s ability to engulf IRBC (paired t-test P = 0·6783). (b) Phagocytosis increased in the AS-exposed short AS-PCt_1/2 (paired t-test P < 0·0001). Phagocytosis decreased in the AS-exposed long AS-PCt_1/2 (P < 0·0001). (c) Regression of AS-induced changes in IRBC phagocytosis and rosetting; F = 15·36, slope significantly non-zero (P = 0·0006). (d) Upon AS-exposure, phagocytosis of purified IRBC increased in both AS-PCt_1/2 groups (paired t-test P <0·0001 for both). (e) In parasite-THP-1 co-culture, the control of short AS-PCt_1/2 experienced increased parasitemia (Friedman test P = 0·0322), but not after AS exposure (P = 0·1574), revealing different growth between both settings (P < 0·0001). In long AS-PCt_1/2, parasitemia increment occurred in AS-exposed setting (P < 0·0001) and control (P = 0·0028), with similar growth (P = 0·5576). (f) THP-1 with engulfed parasite (arrowed); thin smear. (g) AS-exposed short AS-PCt_1/2 co-cultured with THP-1 (arrowed). Most parasites did not develop into rings; thick smear. (h) AS-exposed long AS-PCt_1/2 co-cultured with THP-1 (arrowed), showing abundant ring forms; thick smear. (i) Parasite replication changes between AS-exposed and AS-free settings (mean and S.D. shown) for both AS-PCt_1/2 groups were different (unpaired t-test with Welch's correction P < 0·0001). All images: Giemsa stained; 1000X magnification, scale bar 10 µm. Sample size (n) is stated in each plot.

Fig. 3

How AS mediated rosetting happens. (a) Via Tukey's test, MBCD treatment (membrane cholesterol depletion) did not alter baseline rosetting (P = 0·8546 and 0·9530 for short and long AS-PCt_1/2 respectively). Without MBCD treatment, AS stimulated rosetting, albeit of different extents (P = 0·0064 for short AS-PCt_1/2; P < 0·0001 for long AS-PCt_1/2). After MBCD treatment, rosetting was not altered by AS (P = 0·9915 and 0·7746 for short and long AS-PCt_1/2, respectively). (b) Removal of selective rosetting ligands from the IRBC surface by trypsin to decipher the ligands involved in AS-mediated rosetting. For each of the trypsin groups, Wilcoxon matched pair signed rank tests were performed to compare the rosetting rates obtained from AS-free and AS-exposed settings for each of the enzyme treatment settings. In the non-trypsinised control setting, AS significantly increased rosetting rates (P = 0·0078). For IRBC treated with 10 µg/ml and 1 mg/ml, AS did not exert significant changes to the rosetting rate (P > 0·9999 and P = 0·5000, respectively). (c) AS-mediated rosetting rate changes in P. falciparum isolates with different genotype combinations of K13, PF3D7_0223300 and PF3D7_0200300. Error bars represent mean and S.D. Each combination was represented by a distinct colour. ANOVA with two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli was performed. The AS-mediated rosetting rate changes in K13 mutants with deletions in PF3D7_0200300 and PF3D7_0223300 were significantly higher than those of K13 mutants with deletions in either of the two genes (P = 0·0005), K13 WT with deletions in both genes (P = 0·0186) and K13 WT without deletions in both genes (P < 0·0001). Significant difference was found between K13 mutants with deletions in either of the two genes and K13 WT without deletions in both genes (P = 0·0076), but not with K13 WT with mutations in both genes (P = 0·2646), and K13 WT with deletions in either of the two genes (P = 0·5317). The AS-mediated rosetting rate changes in K13 WT with deletions in both genes were significantly higher than those of K13 WT without deletions in both genes (P = 0·0011). For comparison between K13 WT with deletions in either of the two genes and K13 WT without deletions in both genes, P = 0·0395. (d) Relationship between timeline (red lines; year 2010 as cut-off point, labelled as ‘before_2010’), K13 mutation development (‘K13_status’), deletions at SNP points of genes PF3D7_0200300 (upper left) and PF3D7_0223300 (lower right). Line thickness reflects the effect size of Cramér's V association. From Fisher's with Bonferroni tests, significant associations were found amongst deletions in PF3D7_0200300 and PF3D7_0223300 with K13 mutations (indicated by grey lines) and timeline (indicated by red lines), implying strong linkage disequilibrium. Between timeline and deletion occurrence in PF3D7_0200300, P range = 0·001 – 0·0086. Between timeline and deletion occurrence in PF3D7_0223300, P range = 0·0067 – 0·0069. Within each gene, the occurrence of deletion at different sites were strongly associated with each other. Deletions within PF3D7_0223300 were associated with the occurrence of K13 mutations (range of P = 0·0015 – 0·0029). Statistical details are available in Supplementary Table 5.

Fig. 4

Hypothesized mechanisms of P. falciparum rosetting against ART. (a) Schizont stage-IRBCs with different rosetting phenotypes are exposed to ART. (b) Although not immediately killed by ART directly, the schizont-IRBC that cannot be stimulated to form rosette upon drug exposure will experience increased cellular stress induced by the drug. (c) Following this, the host phagocyte will easily engulf such stressed IRBC, leading to effective parasite clearance. (d) Schizont-IRBC that can form rosette upon the drug exposure has lower chance of being engulfed by the phagocytes, and these rosettable IRBC can resist AS-induced stress better. (e) As a result, these schizonts can complete their schizogony, producing many ring stage-IRBCs that are less susceptible to the drug.