Phenotypic screens reveal Plasmodium falciparum genetic factors associated with infection of sickle-trait cells

Abstract

Background:

Malaria causes over 200 million cases and more than half a million deaths annually. In many African regions, hemoglobinopathies, such as sickle cell trait (HbAS), confer partial protection against severe P. falciparum malaria. HbAS significantly reduces the risk of severe, life-threatening malaria by over 90%. This study aims to describe a new analysis for the piggyBac transposon-based mutagenesis phenotypic screen to identify genes that influence the mechanisms behind this protection and tolerance of P. falciparum to the HbAS intracellular microenvironment, providing insights into potential new targets for malaria intervention and the evolutionary relationship between host and parasite.

Methods:

We optimized and successfully employed a phenotypic screen using a piggyBac transposon-mutant library of P. falciparum to identify genetic factors essential for parasite survival in HbAS RBCs. Parasites were cultured in vitro in HbAS and control HbAA RBCs. Parasite growth was assessed via Quantitative Insertion Site Sequencing (QIseq) to determine sensitivity of each mutant in response to the conditions of HbAS RBCs identifying sensitive and tolerant mutants. Finally, a pairwise comparison was performed between HbAS and previously published piggyBac screens to infer potential links between HbAS infection and parasite responses to heat-shock, antimalarial drugs and oxidative stress.

Results:

Our findings revealed that P. falciparum mutants sensitive to HbAS growth are associated with genes involved in signaling pathways, exported proteins, and host-interaction genes. These genetic factors overlap with those involved in the parasite’s response to oxidative stress and antimalarial drug sensitivity, such as artemisinin derivates and proteasome inhibitor.

Conclusions.

Our study identifies genetic factors influencing P. falciparum infection in HbAS RBCs, shedding light on how HbAS may counteract with the parasite, suggesting a connection between sickle-trait infections and other stress responses, such as heat-shock, artemisinin and oxidative stress.

Introduction

Malaria kills more than half a million people and sickens over 200 million more every year [1]. In many endemic countries of Africa, several hemoglobinopathies confer some resistance to malaria’s severe morbidity and mortality [1]. Children with sickle cell trait hemoglobin (HbAS), representing individuals heterozygous for healthy adult (HbA) and sickle hemoglobin (HbS), have strong protection against Plasmodium falciparum [2]. HbAS limits clinical disease against this most virulent cause of malaria to usually mild symptoms and reduces the risk of severe, life-threatening malaria by >90% [3]. Understanding the mechanisms by which HbAS confers such robust protection could lead to new strategies to combat malaria.

The natural protection against severe P. falciparum malaria conferred by HbAS provides an opportunity to identify critical mechanisms of parasite pathogenesis that are neutralized. Despite the strong epidemiologic evidence of selection for HbAS, the biology and key molecular mechanisms behind the protection against severe malaria are not fully understood. There are several mechanisms linked to the protection offered by HbAS against severe malaria, including acquired host immunity [4], inhibition of parasite translation by host microRNAs [5], parasite death by premature hemolysis [6], dysfunctional hemoglobin digestion [7–9], and defective intra-erythrocytic trafficking of parasite exported proteins with reduced cytoadherence [10–12]. Notably, the phenotype of attenuated cytoadhesion can be phenocopied in normal RBCs by the induction of oxidative stress (RBCs)[13], suggesting that downstream attenuated pathogenesis in part, mediated by the sickle-trait RBC’s redox imbalance [13].

Intraerythrocytic redox balance also mediates the effectiveness of the critical antimalarial artemisinin: we recently showed that the pre-induced oxidative stress in the RBC, mimicking the environment found in hemoglobinopathies, triggers a parasite response that reduces the oxidative toxicity of artemisinin, thereby lowering its effectiveness [14]. This intraerythrocytic oxidative stress, while beneficial in protecting against severe malaria, also plays a role in artemisinin’s mechanism of action and the development of parasite resistance [15, 16]. Reactive oxygen species and superoxide are key to both the efficacy of artemisinin and the parasite resistance mechanisms [16, 17]. Current research suggests that metabolic pathways that enhance fitness upon exposure to artemisinin [18, 19] mirror those which enhance fitness following other experimentally-induced stresses [20, 21]. This raises the possibility that prevalent hemoglobinopathies may alter the fitness landscape for parasites and thereby prime their stress response mechanisms - indeed, parasites infecting people with HbSS have reduced susceptibility to dihydroartemisinin [22, 23].

There is clear evidence that P. falciparum parasites have adapted evolutionarily to sickle hemoglobin: recent genome-wide screens of people with severe malaria have identified 4 parasite loci that are associated with HbS [24, 25]. These adaptations are most clearly associated with the ability of the parasite to overcome HbS-mediated resistance and cause disease, though there may be more subtle adaptations that promote parasite growth and persistence during the blood stage. To further explore more immediate parasite adaptations to HbAS, forward-genetic screens can identify pathways essential for parasite survival in sickle-trait RBCs. Pooled genetic screens, which allow rapid and high-throughput identification of genes influencing specific phenotypes, have been successfully applied in apicomplexan species [20, 26–30]. In this study, we optimized and employed a phenotypic screening strategy using a library of isogenic P. falciparum piggyBac-mutant parasites – referred to as the pilot piggyBac library [20, 29–32] – to identify genetic factors associated with the parasite’s adaptation to sickle-trait cells. As previously established, under a phenotypic selection such as the HbAS microenvironment, piggyBac mutants with mutations in genes required for survival under phenotype conditions will exhibit impaired growth compared to mutations in genes not essential for survival. Our findings reveal that the genes associated with parasite survival in HbAS RBCs are also correlated to oxidative stress and antimalarial responses. These insights enhance our understanding of host-parasite interactions and highlight critical pathways involved in parasite survival in hemoglobinopathic environments, providing new avenues for future research to identify potentially druggble weaknesses in the parasite’s metabolism.

Results:

Sickle-trait pilot piggybac screen phenotype identification

The ‘pilot’ piggybac library is a well-characterized collection of 128 P. falciparum piggyBac-mutant clones representing disruptions in genes across various functional categories [33]. Here, we conducted two biological replicates of screens of the pilot piggyBac library in HbAS (sickle-trait) and HbAA (normal) RBCs for three and six cycles (Figure 1A) for two different HbAS donors. As previously established, mutants with disruptions in genes critical to survival under the tested experimental condition, or phenotype selection, are expected to exhibit reduced growth compared to mutants in genes not linked in the response [20, 30]. Mutant growth phenotypes were identified and differentiated using QIseq (Methods –Supplementary material Table S1 and S2). While QIseq normalized read counts showed high correlation within timepoints between biological replicates, we observed lower correlation between those at 6 and 3 cycles (Supplementary Figure S1). Sensitive mutants present disruptions in genes essential for growth, showed lower relative QIseq read counts (Log₂ fold-change HbAS/HbAA), indicating a role for these genes in the survival response to sickle-trait cells. Crucially, phenotypic interpretation of piggyBac screens is primarily focused on identifying sensitive mutants [30], as these typically exhibit reduced growth in HbAS compared to HbAA, often indicating disruption of genes essential for parasite survival under sickle-trait stress conditions. In contrast, mutants classified as tolerant—those showing relatively better growth in HbAS—may not necessarily reflect genes directly involved in adaptation to the HbAS environment. Their apparent enrichment can result from inherent competitive advantages under standard (HbAA) culture conditions, rather than true fitness gains in HbAS. This phenomenon, previously described in pooled screens [31, 33], can lead to misinterpretation of tolerant phenotypes due to skewed relative abundances within the mixed mutant population.

Figure 1:

A) Approach for the sickle cell trait (HbAS) piggyBac (pB) phenotype screen. The pilot pB library (n = 128 mutants) was cultured in sickle trait RBCs HbAS (two donors). Control flasks (HbAA ‘normal’ RBCs) were cultured continuously in parallel. After 3 cycles growth, then again after 6 cycles growth, the cultures were harvested, and insertion sites were identified by QIseq (see Methods). B) Pearson correlations between relative growth in growth control conditions (Log₂FC T1/T0 and Log₂FC T2/T1), experimental conditions (Log₂FC HbAS/HbAA), and between control and experimental conditions. C) Pearson correlations between relative growth (Log₂FC HbAS/HbAA) of sensitive and tolerant piggyBac mutants at 3 and 6 cycles and the P. falciparum 3D7 cultured in HbAS transcriptome data (Log₂FC HbAS/HbAA) from Saelen et al 2021. Y axes indicate the Pearson correlation coefficient (r) obtained for each timepoint (X axis) (Supplementary Material). Expression trend plots were generated using GraphPad with LOESS smoothing.

Importantly, piggyBac mutants exhibit inherent competition when grown in a pooled library; mutants with higher fitness under ideal culture conditions tend to outcompete others and become overrepresented in the library [33]. To account for this bias, we performed Pearson correlation analyses using the relative growth observed in control cultures—specifically, Log2 fold-change of HbAA_T1/HbAA_T0 (growth control at three cycles) and HbAA_T2/HbAA_T0 (growth control at six cycles). These values showed weak correlation with relative growth under HbAS conditions (Log2 fold-change HbAS/HbAA at three and six cycles) (Figure 1B). This weak correlation between growth in sickle-trait vs. control RBCs supports that phenotypes reflect specific responses to the HbAS environment rather than general differences in growth capacity, with phenotypes exhibiting lower correlations between experimental vs. control conditions considered the most confident [20, 30].

Of relevance, a search on the PlasmoDB database (https://plasmodb.org/plasmo/app/search/snp/NgsSnpsByGeneIds) using gene IDs revealed that 56 out of 106 genes identified as sensitive in our screen harbor single nucleotide polymorphisms (SNPs) with >2% minor allele frequency in P. falciparum field isolates. These polymorphisms were found in samples from multiple African countries (Democratic Republic of Congo, Gambia, Ghana, Kenya, Mali, Mozambique, Senegal, Sudan, Togo, and Uganda) with high HbS prevalence (Table S3).

To correlate mutant phenotypes with gene expression during asexual blood-stages development, we compared genes identified as sensitive and tolerant genes at 3 and 6 cycles with a transcriptome dataset of developmental time points from P. falciparum infections in HbAS erythrocytes, using a previously published study with extensive time-point data [34]. The genes with increased sensitivity to sickle-trait cells growth exhibited lower expression during most of the intraerythrocytic cycle, particularly from the mid-late ring to trophozoite stages. At the schizont stage, their expression was higher compared to genes associated with decreased sensitivity (tolerant genes) (Figure 1C).

P. falciparum genes with increased sensitivity to survival in sickle-trait

Sensitive phenotypes are closely linked to the essentiality of parasite survival in sickle-trait cells. Our results demonstrate that, at three cycles, genes exhibiting increased sensitivity to growth in sickle-trait cells are associated with exported proteins (such as serine/threonine kinases, including protein serine/threonine kinase FIKK9.3 - PF3D7_0902200) and host-interaction genes (such as inner membrane complex protein 1b, IMC1b - PF3D7_1141900). Signaling pathway phosphatases also exhibit sensitivity at both 3 of growth, including mitogen-activated protein kinase phosphatase 1 (MKP1 - PF3D7_1305500) at 3 cycles. Interestingly, genes encoding proteins associated with artemisinin sensitivity, such as Kelch13 compartment-localized KIC1 (PF3D7_0606000) [35] and dynein heavy-chain (DHC - PF3D7_1122900), which is sensitive to DHA [14, 30], also show increased sensitivity to growth in sickle-trait cells at three cycles (Figure 2A–B). The Endoplasmatic reticulum (ER) membrane protein complex subunit 3 (EMC3 - PF3D7_1360200), a gene associated with reduced gametocyte production [29], is also sensitive at both 3 and 6 cycles. Gene ontology enrichment analysis of sensitive genes revealed organelle-related processes, such as “bounding membrane of organelle” (GO:0098588) at 3 cycles and “Apicoplast” (GO:0020011) at 6 cycles. For instance, acyl-CoA synthetase 5 (ACS5 - PF3D7_0731600), an apicoplast-localized protein-coding gene, is also involved in host interactions within knobs [36]. Intriguingly, exported proteins, apicoplast and phosphatases were also encountered in previous genome-wide studies [24, 25] that identified variations in four regions of the P. falciparum genome enriched in parasites infecting individuals with sickle hemoglobin. Particularly, these regions encode two exported proteins (serine/threonine kinase FIKK4.2 - PF3D7_0424700 [25] and PF3D7_0220300 [24]), a tyrosine phosphatase (PF3D7_1127000) [24], and an acyl-CoA synthetase (ACS8 - PF3D7_0215300) [24], which is an apicoplast-localized gene [37, 38] that also contains a Plasmodium export element (PEXEL)-like motif [39].

Figure 2:

A) Genetic factors associated with sickle-trait (HbAS) RBCs infections identified in the piggyBac phenotypic screen. Relative differentiation of each piggyBac mutant in the library was determined by ranking mutants from low to high. The genes ranked in the lowest quartile are designated sensitive (indicated in red). The top-ranked quartile of genes is designated tolerant (indicated in blue). * Indicates mutant sensitive in one donor, ** indicates mutant sensitive in both donors. Volcano plots showing the assigned phenotypes for each donor separately at each timepoint is provided in Figure S2. The entire piggyBac screen dataset is provided in Supplementary Material (Table S2). C) Functional enrichment of significant gene ontology (GO) terms for sensitive piggyBac mutants at 3 and 6 cycles growth vs all other mutants in the library. Terms above the dotted line are significant (p. value < 0.05, Fisher/elim-hybrid test). The entire GO-dataset is provided in Supplementary Material (Table S4).

We then performed HbAS growth assays on individual mutant clones for three cycles (see Methods) to validate piggyBac mutant phenotypes from pooled screening. We selected mutants of each phenotype for validation: three sensitive mutants (EMC3pB, KIC1 pB and IMCb pB), two tolerant mutants (PfEMP1 pB - erythrocyte membrane protein 1 PF3D7_0808700, and SF3A2 pB – splicing factor 3A2 PF3D7_0619900) and two neutral phenotypes (protein phosphatase PF3D7_0615900 and LRR5 pB - PF3D7_1432400). Additionally, we included the wild-type NF54 line in the phenotype validation experiments as a comparator to the parental strain baseline phenotype. Note that this parental line is does not carry an integrated transposon and therefore is absent from the piggyBac pool libraries. First, to validate the screens, we compared the piggyBac clones among themselves, given that the library screen phenotypes rely on the ranked relative abundance of each mutant (see Methods). We observed consistent phenotypic trends across the library screens and clones (Figure 3), except for one tolerant mutant (PfEMP1pB). Specifically, the two sensitive mutants had a significantly lower parasitemia ratio (HbAS/HbAA) compared to at least one of the two tolerant mutants. Second, we compared the piggyBac clones with wild-type NF54. All sensitive clones showed a lower parasitemia ratio compared to wild-type NF54, though only one rose to statistical significance (EMC3pB; p < 0.001, one-way ANOVA).

Figure 3: Individual phenotypic screens of piggyBac mutant clones in sickle-trait (HbAS) RBC.

Seven piggyBac mutant clones and Wild-type NF54 were cultured parallelly HbAS and in HbAA control RBCs for three consecutive cycles. The cumulative parasitemia was determined for each condition, and the growth ratio (HbAS/HbAA) was calculated to categorize phenotype. The assay was conducted in three independent biological replicates. Comparisons among parasite lines were analyzed using one-way ANOVA followed by Tukey’s test (*p<0.01, **p<0.001, ***p<0.0001, ****p<0.00001).

Factors linked to adaptation of parasites to different stress environments

Our prior piggyBac mutant screens have identified overlapping gene sensitivities to related phenotypes, such as fever survival, heat-shock response, and artemisinin susceptibility [20]. These similarities extended to pathways implicated in artemisinin resistance, including responses to protein damage and oxidative stress [18, 40, 41]. Given that similar phenotypes are observed in parasite infected sickle-trait RBCs with and pre-treated oxidative stress RBCs [13], we took advantage of our previously published parallel phenotypic screens of the same piggyBac mutant pilot library [20] to investigate potential relationships between the sickle cell trait and other stress responses. These included HS, exposure to two artemisinin derivatives-dihydroartemisinin (DHA) and artesunate (AS) - increased oxidative stress RBCs (two time points T1, 3-cycles, T2, 6-cycles), and exposure to a proteasome inhibitor, bortezomib (BTZ).

Our analysis revealed that the blood-stage growth phenotypes of piggyBac mutants were highly correlated following exposure to sickle-trait RBCs after 3 cycles and exposure to either artemisinin derivative (DHA and AS), oxidative stress, HS, or BTZ (Figure 3A). Specifically, for oxidative stress, a significant correlation was observed with oxidative stress sensitivity at six cycles (oxi-6cyc T2), similar to the previously reported correlation between HS and oxidative stress sensitivity at six cycles [20].

Genes with significantly increased sensitivity to oxidative stress (oxi-6cyc) and sickle-trait conditions at three cycles included: three conserved Plasmodium genes encoding putative proteins of unknown function, two PfEMP1 genes, IMC1b, DHC, and FIKK9.3 (Figure 4B and supplementary data). The shared sensitivity of several Plasmodium genes across these conditions highlights potential molecular links between the sickle cell trait and artemisinin susceptibility. Notably, genes such as DHC, KIC1, and EMC3 may play key roles in mediating parasite survival under these stress conditions. Interestingly, EMC3, an endoplasmic reticulum protein [42, 43] and a gametocyte hypo-producer gene [29], displayed increased sensitivity to all screened stressors except oxi-6cyc.

Figure 4:

A) Pairwise correlation between mutant phenotypes (Log2FC) in three (HbAS_3cyc) and six cycles (HbAS_6cyc) growth in sickle-trait RBCs (HbAS) and previously published pB phenotype screens (Log2FC) (Zhang 2021) of the same pilot piggyBac library used in the current study. Prior screened conditions included heatshock (HS); two artemisinin derivatives in two sublethal concentrations (dihydroartemisinin, DHA; artesunate, AS at IC10 and IC25); oxidative stress-induced RBCs at 3 and 6 cycles growth (OXI-3cyc, OXI_6cyc respectively), and exposure to a proteasome inhibitor, bortezomib (BTZ_IC10). The control screen was also included in the comparison. HbAS_3cyc phenotypes showed significantly higher correlation with all other stressors than HbAS_6cyc phenotypes. B) Scatter plot comparing HbAS_3cyc phenotypes (Log2FC) with each significant correlated phenotype: HS, BTZ_IC10, AS and DHA_IC15 and OXI-6cyc (Log2FC). Genes of interest sharing sensitive phenotypes in both screens are listed in green on each plot. Point colors indicate the phenotypes assigned in Zhang et al 2021.

Discussion

HbAS consistently protects against severe P. falciparum malaria and consequently the parasite has adapted to sickle hemoglobin via multiple genetic mutations. To understand specific parasite genetic factors that enable parasite survival under various other conditions, such as heat shock, antimalarials, and oxidative stress [20, 29, 30], we have previously used P. falciparum piggyBac mutagenesis and the robust QIseq analysis. In this current study, we applied this approach to the RBC hemoglobinopathy prevalent in African malaria-endemic populations, demonstrating that the piggyBac screening approach can identify genes associated with parasite survival in sickle-trait cells. By exploring results from previously published screens [20], we could distinguish genes associated with sickle-cell survival from those involved in response to other conditions, such as fever, drug exposure, and oxidative stress. This comparison allowed us to reveal the overlap in parasite mechanisms of adaptation to different stress conditions.

Our findings reveal that signaling pathways, exported proteins, and host-interaction genes are associated with parasite adaptation to sickle hemoglobin. These results are consistent with genome-wide studies of parasites infecting people with sickle hemoglobin, which identified variations in three regions of the parasite genome enriched in parasites infecting Gambian and Kenyan children with sickle hemoglobin [24]. These loci encode acyl-CoA synthetase (ACS8 - PF3D7_0215300), an apicoplast protein [38]; a putative tyrosine phosphatase of the food vacuole (PF3D7_1127000) [44]; as well as exported proteins PF3D7_0220300 and FIKK 4.2 (PF3D7_0424700) [25]. In our current screen, piggyBac mutants for ACS-5 (a knob component [36], and annotated as apicoplast gene), FIKK 9.3 and MPK1 phosphatase have decreased growth in sickle-trait RBCs, indicating that these exported parasite proteins enable parasite survival in the HbAS RBC microenvironment. In HbAS RBCs, parasite-induced actin reorganization, knob formation, and Maurer’s clefts development are impaired, reducing protein export and cytoadherence, which correlates with protection against severe malaria [10, 12, 45, 46]. Furthermore, parasite infection and sickle hemoglobin increase intraerythrocytic oxidative stress [13], altering the phosphorylation of parasite proteins [47, 48], which affects parasite development, cytoadherence, membrane channel activities, and mechanical properties. Changes in kinase/phosphatase balance influence these processes, highlighting phosphorylation as a key mechanism in malaria pathogenesis and the protective role of sickle hemoglobin [49]. Additionally, previous study has demonstrated that the parasite’s overall transcriptional program remains largely unchanged in infected HbAS RBCs. Most of the differentially expressed genes in the HbAS condition appear late in the asexual intraerythrocytic development cycle [34], which could explain the tendency of sensitives genes be highly expressed in late stages (Figure 1C).

The disruption of EMC3 rendered parasites very sensitive to growth in sickle-trait RBCs collected from three separate donors. Although EMC3 has not been extensively characterized in P. falciparum, studies in other eukaryotes indicate that it is a core component of the EMC, a conserved complex involved in the insertion, folding, and stabilization of multi-pass transmembrane proteins into the ER membrane [42, 43], including formation of phospholipid membrane [50, 51]. In P. falciparum. the EMC is predicted to play a critical role in supporting the parasite ER membrane systems required for protein trafficking, and survival within the host erythrocyte. Under HbAS (sickle-trait) conditions, the parasite encounters a more oxidative and metabolically changed environment and altered protein trafficking [11, 12] which likely increases the burden on ER function and membrane integrity. We hypothesize that disruption of EMC3 compromises the parasite’s ability to tolerate this host-induced stress, particularly by impairing the formation of phospholipid membrane and the export proteins essential for survival in HbAS RBC. The consistent and significant growth sensitivity of the EMC3pB mutant under this HbAS environment highlights EMC3 as a candidate for further functional characterization and potential therapeutic targeting.

The pre-induced intraerythrocytic oxidative stress caused similar morphological effects as observed in hemoglobinopathic RBCs [13]. The unfavorable intraerythrocytic conditions disrupts actin remodeling, inhibit actin polymerization and protein export in P. falciparum-infected erythrocytes leading to abnormal Maurer’s clefts and knob formation, diminishing functional cytoadherence of these infected RBCs, which have been suggested to contribute to protection against severe malaria [13]. Concomitantly, hemoglobinopathic RBCs can lead to parasite responses that counterbalance artemisinin toxicity by enabling parasite tolerance of the reactive oxygen species generated by artemisinin, which is its primary mechanism of action [15, 16]. A transcriptome study [34] found that HbAS alter parasite transcription linked to oxidative stress response, protein export machinery and protein folding machinery, and proteasome and oxidative stress pathways have also been found to be upregulated in artemisinin-resistant parasites [19, 41], suggesting that hemoglobinopathies may alter the parasite fitness landscape and select for adaptations that enhance resistance to artemisinin compounds. Our forward-genetic phenotypic screens identified genetic factors that regulate P. falciparum sensitivity to sickle-trait cells, correlating with the parasite responses to malarial fever, oxidative stresses, artemisinins and proteasome inhibitor (Figure 4). Importantly, these findings are consistent with analysis of recent parasites field isolates wherein exported proteins (FIKK family) and components of organellar metabolism (such as ACSs), are critical to altering sensitivity to artemisinin [19, 52, 53], reinforcing that the malaria parasite may adapt a conserved response to different stressors. The relevant presence of SNPs in several of sensitive genes (Table S3) also indicate parasite adaptation in response to the HbAS intracellular environment.

In conclusion, even in a small-scale phenotypic screen our findings reveal genetic factors involved in the P. falciparum response to HbAS RBCs infection, highlighting genetic factors by which HbAS may neutralize the parasite. The vulnerable pathways in the piggyBac mutant perturbations that are deleterious to the malaria parasite in HbAS infections provide new targets to combat parasite mediators of disease and suggest a link between sickle-trait cell infections and other stress responses. Exploring the characteristics of the host-parasite evolutionary relationship offers new insights into adaptation to different stressors through metabolic changes, specifically in oxidative stress, phosphorylation and signaling, and exported proteins.

Methods

Parasite culture and maintenance

The P. falciparum piggyBac-mutant library was cultured in 4% hematocrit (O+ erythrocytes from Interstate blood bank, Memphis, TN) and 0.5% Albumax II in RPMI 1640 medium (Invitrogen) supplemented with 50 μ g/ml hypoxanthine (Sigma) and 25 mM HEPES (Invitrogen). The culture flasks were grown in an incubator with continuous flow of mixed gas (90% Nitrogen, 5% CO2 and 1% O2 respectively). HbAA and HbAS RBCs were collected from donors after written informed consent (Duke IRB# Pro00007816), with genotypes confirmed by Sanger sequencing.

Generation of piggyBac mutant pilot-library

The pilot library has been previously applied for multiple phenotypic screens in P. falciparum. It contains 128 isogenic pB mutants that were created during our whole-genome random mutagenesis saturation project [31, 33]. Briefly, the piggyBac mutant parasite clones were thawed individually and grown to a 1–2% parasitemia in T25-flasks. Aliquots of the pilot library were generated by combining equal volumes of all clones and cryopreserving about 100 vials according to standard methods, ensuring a supply of enough biological replicate samples for use in phenotypic screens.

Phenotype screen: growth in sickle-trait cells

Phenotypic screens of parasite growth in sickle-trait cells were performed using a 128 piggyBac-mutant library (pilot piggyBac library). The piggyBac-screening analysis pipeline used before [20, 30] was adapted for this current study. Briefly, two aliquots (two biological replicates) of the piggyBac -library were independently thawed, growth in HbAA for two or three cycles growth to reach enough parasitemia to then split (P=0.5%) into three 10mL flasks: two experimental flasks HbAS donor S02 and HbAS donor S06, and one as growth-control flask with HbAA cells. Experimental and control flasks were maintained in parallel at 37°C to minimize potential batch effects. A time point zero sample, T0, was immediately harvested. Parasites of each groups were harvested immediately after three cycles-growth (T1), then again after an additional three growth-cycles (T2) for gDNA-extraction and phenotype-analysis by QIseq [32]. Important to highlight, the piggyBac library was not synchronized. The synchronization can cause loss of mutants with lower relative fitness within the library and the screen results can be mis-interpretated. The pB-mutants in the library present dysfunction growth, with inherent competition within the pool [20, 31, 32].

The QIseq original read counts and normalized reads are provided in Supplementary material table S1.

Phenotype identification

Quantitative insertion-site sequencing (QIseq) was performed as previously described to quantify the relative abundance of each piggyBac mutant in the screen pools [33]. The piggyBac screen analysis pipeline established in previous studies [20, 30] was adapted for this study. Briefly, raw read counts per insertion site were normalized, and fold changes (HbAS/HbAA) for each mutant were calculated separately for each biological replicate, time point (T1–3cycles and T2–6cycles), and donor (S02, S06) using the DESeq2 package [54], as used before in previously phenotypic screens (Pires, . Log2 fold changes HbAS/HbAA were then ranked from lowest to highest within each time point and donor. Mutants falling within the lower and upper quartiles of the Log2 fold-change HbAS/HbAA distribution were classified as sensitive or tolerant to sickle-trait RBC growth conditions, respectively. Significant phenotypes were defined as those within these quartiles that also showed a p-value < 0.05 at 3 cycles and an adjusted p-value < 0.05 at 6 cycles. For data visualization and Gene Ontology (GO) analysis, the average Log2 fold changes HbAS/HbAA across the two donors were calculated, ranked, and plotted in Figure 2. The complete phenotype dataset, including normalized read counts, fold changes, p-values, and assigned phenotypes, is provided in Supplementary Table S2.

Gene ontology (GO) enrichment

GO-enrichment analyses were performed by testing GO-terms mapped to the sickle-trait phenotypic categories of interest against a background of GO-terms mapped to all other genes using our R package pfGO [55] (v 1.1). Categories of interest were sensitive 3 cycles (sensitive 3 cycles only), sensitive 6 cycles (sensitive 6 cycles only), sensitive 3 and 6 cycles (sensitive 3 and cycles), tolerant 3 cycles (tolerant 3 cycles only), tolerant 6 cycles (tolerant 6 cycles only), tolerant 3 and 6 cycles (tolerant 3 and 6 cycles), and neutral (Figure 2). The GO-term database was created from the latest curated P. falciparum ontology available at the time of analysis from PlasmoDB [56] and enrichment was assessed via a weighted Fisher/elim-hybrid p <=0.05 (v. 57). The fraction of genes represents the number of significant genes annotated to a given GO-term in each of the categories divided by the total number of genes annotated to that GO-term included in the analysis for all categories (background-set). The entire GO-data set is provided in Supplementary material table S4.

Individual Phenotypic Screens of piggyBac Mutants in Sickle-Trait (HbAS) RBC

Selected piggyBac mutants representing distinct phenotypes (Figure 3) from the pooled library screen were thawed and initially cultured in HbAA (normal) RBCs. Parasite lines were synchronized at the schizont stage using a Percoll gradient. Synchronized schizonts from each line were then equally divided into two flasks: one containing HbAS RBCs and the other containing HbAA RBCs (growth control). Cultures were maintained until reinvasion. At the ring stage (~16 hours post-synchronization), parasitemia was adjusted to 0.5% at 2% hematocrit, and cultures were maintained for three consecutive asexual replication cycles. Media were replaced daily, and parasitemia was measured every 48 hours by staining ~30 μL of culture with Hoechst 33342 (1:5000) diluted in phosphate-buffered saline (PBS), incubated at 37°C for 10 minutes, followed by two PBS washes. Percent parasitemia was quantified by acquiring 80,000 events per sample using a Miltenyi MacsQuant Flow Cytometer. Giemsa-stained smears were also prepared to confirm parasitemia and assess parasite morphology and viability. HbAA and HbAS conditions were maintained in parallel throughout the assay. After three cycles, cumulative parasitemia was determined for each condition, and the growth ratio (HbAS/HbAA) was calculated. Statistical analyses were performed using GraphPad Prism (Version 10.4.1). The assay was conducted in three independent biological replicates.

Data availability.

Raw QIseq data sets generated for this study were deposited to the European Nucleotide Archive under study accession code: ERS8537951, ERS8537948, ERS8537943, ERS8537949, ERS8537942, ERS8537946, ERS8537950, ERS8537944, ERS8537940, ERS8537947, ERS8537945, ERS8537941, ERS8537953, ERS8537952. Sample accession codes with their descriptions (samples ID) and processed QIseq data are provided in Supplementary material.