Plasmodium falciparum Maf1 Confers Survival upon Amino Acid Starvation

ABSTRACT The target of rapamycin complex 1 (TORC1) pathway is a highly conserved signaling pathway across eukaryotes that integrates nutrient and stress signals to regulate the cellular growth rate and the transition into and maintenance of dormancy. The majority of the pathway’s components, including the central TOR kinase, have been lost in the apicomplexan lineage, and it is unknown how these organisms detect and respond to nutrient starvation in its absence. Plasmodium falciparum encodes a putative ortholog of the RNA polymerase (Pol) III repressor Maf1, which has been demonstrated to modulate Pol III transcription in a TOR-dependent manner in a number of organisms. Here, we investigate the role of P. falciparum Maf1 (PfMaf1) in regulating RNA Pol III expression under conditions of nutrient starvation and other stresses. Using a transposon insertion mutant with an altered Maf1 expression profile, we demonstrated that proper Maf1 expression is necessary for survival of the dormancy-like state induced by prolonged amino acid starvation and is needed for full recovery from other stresses that slow or stall the parasite cell cycle. This Maf1 mutant is defective in the downregulation of pre-tRNA synthesis under nutrient-limiting conditions, indicating that the function of Maf1 as a stress-responsive regulator of structural RNA transcription is conserved in P. falciparum. Recent work has demonstrated that parasites carrying artemisinin-resistant K13 alleles display an enhanced ability to recover from drug-induced growth retardation. We show that one such artemisinin-resistant line displays greater regulation of pre-tRNA expression and higher survival upon prolonged amino acid starvation, suggesting that overlapping, PfMaf1-associated pathways may regulate growth recovery from both artemisinin treatment and amino acid starvation.

posed the so-called "sleeping beauty" hypothesis: ring-stage resistant parasites enter a state of dormancy, arresting the cell cycle and decreasing metabolic activity to limit damage, and then resume growth after drug concentrations have decayed below effective levels (3)(4)(5)(6)(7)(8). However, more-recent work using field isolates known to carry mutations conferring the delayed clearance phenotype has cast doubt on the model of full cell cycle arrest and a transition to dormancy. Instead, studies have suggested that artemisinin derivatives induce a transient slowing of the cell cycle (9) and that parasites carrying resistance-conferring mutations are able to survive this retardation and resume growth, while susceptible parasites are not able to do so (10). The mechanisms by which the drug transiently slows the cell cycle and how the resistant alleles allow recovery remain unclear.
The yeast Saccharomyces cerevisiae has long been a model for the study of induction and maintenance of dormancy. When nutrients are depleted in the stationary phase, cells appear to cease progress through the mitotic cycle and enter a state of low metabolic activity and protein synthesis. This state serves a protective function, as yeast deprived of carbohydrates can remain viable and recover after more than 100 days of starvation (11).
Asexual stages of Plasmodium falciparum enter a state reminiscent of the yeast stationary phase upon nutrient limitation in vitro. When deprived of extracellular isoleucine (Ile), the only amino acid that cannot be obtained from the digestion of hemoglobin, the parasite dramatically slows its cell cycle and exhibits decreased protein synthesis and metabolic activity (12). The parasite can remain in this state for several days and resume normal growth upon isoleucine resupplementation with little loss in viability (12). To date, no parasite genes or pathways have been implicated in the attainment or maintenance of this dormancy-like state.
A likely candidate to govern these responses is the target of rapamycin complex 1 (TORC1) pathway, as it is known to integrate a range of positive and negative growth signals, most notably the presence of amino acids, to drive or inhibit cellular growth (13). The TORC1 pathway is highly conserved throughout eukaryotes and was likely a central signaling hub in the last eukaryote common ancestor (14,15). In mammalian cells in culture, inhibition of mTOR with the drug sirolimus induces G 1 arrest in certain cell types but only slows cell cycle progression in others (16). In yeast, complete chemical or genetic inhibition of TOR drives cells into dormancy or quiescence (17), while partial inhibition slows cell cycle progression (18). A genome-wide deletion screen in yeast found TORC1 pathway components to be the top hits for the ability to maintain viability in stationary phase during prolonged starvation (18).
The majority of the most familiar components of the TORC1 pathway, notably the TOR kinase itself, have been lost through genomic reduction in the evolution of the apicomplexan lineage ( Fig. 1) (14,15). While P. falciparum encodes none of the core TORC1 components, several highly conserved genes that have been peripherally associated with various branches of TORC1 remain. Intriguingly, two such components were recently associated with artemisinin resistance in parasites carrying K13 resistance alleles. Increased levels and activity of both the P. falciparum phosphatidylinositol 3-kinase gene (PfPI3K) (PF3D7_0515300) and PfPKB (also known as PfAkt) (PF3D7_1246900) were shown to confer increased survival of dihydroartemisinin (DHA) treatment in vitro (19), suggesting that these genes may actively regulate the slow growth and recovery program required to survive artemisinin treatment in the absence of TORC1.
A more recently appreciated factor associated with the TORC1 pathway is the RNA polymerase (Pol) III regulator Maf1. Under nutrient-replete conditions, TOR signaling maintains Maf1 in a phosphorylated and inactive state (20)(21)(22)(23). Under starvation conditions, inhibition of TORC1 leads to Maf1 dephosphorylation, allowing it to bind to the RNA Pol III holoenzyme inhibiting Pol III-dependent transcription of tRNAs, the 5S RNA, and other structural RNAs (24,25). Yeast Maf1 null mutants die under conditions of nutrient limitation or when other factors inhibit TORC1, due to inability to regulate Pol III transcription (22,(26)(27)(28). In yeast, Maf1 is one of the most important genes for maintaining viability during long-term starvation in stationary phase (18,29). A putative Maf1 ortholog appears to be conserved in the Plasmodium genus.
Here, we show that the P. falciparum gene is a functional ortholog of yeast Maf1. A transposon insertion line of P. falciparum with defective Maf1 expression is unable to regulate Pol III activity or to maintain viability during the dormancy-like state induced by isoleucine starvation. This mutant displays additional growth and recovery defects for a range of growth-inhibiting forms of stress. Furthermore, an artemisinin-resistant isolate displays more effective Pol III regulation and increased survival upon amino acid starvation, suggesting that PfMaf1 remains a downstream effector of growth regulation pathways in Plasmodium falciparum despite its loss of TORC1. . Plasmodium spp. lack a class I PI3K enzyme and the other components (PTEN, PDK, an Akt homolog containing a PH domain) typically associated with this signaling cascade. Plasmodium spp. do encode a class III PI3K enzyme (PF3D7_0515300) whose ortholog has been implicated in TORC1 signaling in human cells, as well as a PH domain lacking PKB family kinase (PF3D7_1246900) resembling human S6K. The genomes of Plasmodium parasites also encode an apparent ortholog of the TORC1-dependent RNA polymerase III regulator Maf1 (PF3D7_0416500).
PfMaf1 Confers Survival upon Amino Acid Starvation ® indicating that the core region of the P. falciparum ortholog is capable of functionally complementing yeast knockout cells.
The PB-11 parasite clone carries a transposon insertion upstream of the Maf1 open reading frame (ORF). We attempted to disrupt the Maf1 ortholog PF3D7_0416500 with several different plasmids for single-and double-crossover recombination, including multiple attempts using clustered regularly interspaced short palindromic repeat (CRISPR)-CRISPR-associated protein-9 nuclease (CRISPR-Cas9)/guide RNA. None of our efforts was successful, raising the possibility that this gene is essential for intraerythrocytic growth. We were not successful in integrating any C-terminal epitope tags or destabilization domains (no 3= integration was ever detected by PCR in transfected parasites), suggesting that an unmodified C terminus may also be necessary for proper function.
Surprisingly, we were also unable to express ectopic copies of Maf1 in wild-type parasites, using either episomal plasmids or chromosomal integration via the attB/attP system (30). We attempted the use of both N-terminal and C-terminal tags, as well as both strong (eEF1␣) (31) and weak (mRPL2) (32) promoters, but no stable transfectants could be obtained. Unrelated control plasmids routinely produced stable transfectants. Only a single ectopic expression construct produced stable parasites, and it consisted of the P. falciparum Maf1 coding sequence fused to a "DD24" N-terminal FKBP12 destabilization domain (33) expressed from the weak mRPL2 promoter. The resulting line expressed the Maf1 transgene well below endogenous levels and was unresponsive to protein stabilization with the ligand Shield, limiting its utility in any further analyses (Fig. S2). These failed efforts suggest that Maf1 may be toxic to cells if expressed at the wrong time and/or level.
In their efforts to generate a piggyBac mutant collection, Balu et al. generated a Maf1 transposon insertion mutant in the NF54 parasite background (34). In this clone, identified as PB-11, Balu et al. reported the piggyBac transposon to be inserted at the TTAA sequence ending at position ϩ8 of the ORF, which should disrupt the coding sequence to generate a genetic knockout (Fig. 3A).
We obtained this mutant and discovered that the insertion site had been incorrectly assigned. Instead of the ϩ8 position of the ORF, the insertion is at a TTAA sequence To rule out the possibility of additional, unknown piggyBac insertions in the genome, we aligned one file of the paired reads to the transposon sequence and the pair file to the genome (and vice versa) to see where each transposon-aligning paired read mapped (35). Paired ends were found to align only to the Maf1 locus or to regions of the genome used as regulatory sequences within the transposon itself (such as the calmodulin promoter or the HrpII 3= untranscribed region [3=UTR]) (Table S1). Additionally, we analyzed the PB-11 genome for single nucleotide polymorphisms (SNPs) and indels within all annotated open reading frames in the genome but found none of significance compared to the parental line or a sister piggyBac insertion mutant sequenced previously (35) (Table S2). The latter results support the notion that any phenotype observed in the PB-11 mutant line is a consequence of the transposon insertion.
The PB-11 mutant displays an altered Maf1 expression profile. The Ϫ53 insertion in the PB-11 line does not ablate expression of Maf1. The Maf1 transcript is detectable in PB-11 mutants throughout the asexual life cycle but does not match the wild-type expression pattern ( Fig. 4) (Fig. S3). In later stages of the cell cycle, when wild-type Maf1 mRNA abundance dramatically increases, expression of mutant Maf1 stays at early-cycle levels.
The insertion of an~3 kb transposon at 53 bp upstream of the start codon would normally be expected to disrupt the activity of the native promoter sequence. Therefore, we performed 5= rapid amplification of cDNA ends (5=-RACE) on Maf1 mRNA from both wild-type and PB-11 mutant parasites. In the wild-type cells, we mapped the transcription start site to 93 bp upstream of the Maf1 start codon, meaning that piggyBac inserted into the 5=UTR in the PB-11 mutant (Fig. 3A). In PB-11 cells, the Maf1 transcription start site mapped to 395 bp upstream of the start codon, producing a chimeric 5=UTR consisting of 346 bp of the right arm of the piggyBac transposon (Fig. 3A). In constructing the transposon pXL-BacII-DHFR (pXL-BacII-dihydrofolate reductase), Balu et al. used the full intergenic region between the start codons of the gene encoding calmodulin (PF3D7_1434200) and its opposite gene (PF3D7_1434300) to drive expression of the human DHFR (hDHFR) selectable marker (36). It appears that this region has enough promoter activity to drive transcription of the chimeric Maf1 profile. An analysis of the time course of Maf1 mRNA expression was performed using qRT-PCR at seven time points across the intraerythrocytic cycle and synchronous wild-type (NF54) and mutant (PB-11) parasites. Maf1 expression was quantified relative to that of the seryl-tRNA ligase transcript (PF3D7_0717700). Points represent individual biological replicates (three in total), and curves represent LOESS smoothed models fitted to the data, with the 95% confidence interval indicated by shading. ΔΔCT, threshold cycle method; hpi, hour post-red blood cell (RBC) invasion. mRNA in PB-11 in the direction opposite that of the hDHFR gene. This promoter has been previously reported to be bidirectional when used in plasmids (37).
Antisera raised against full-length Plasmodium berghei Maf1 detected accumulation of Maf1 protein in both wild-type and mutant cells. The Maf1 product is a lowabundance protein and was largely undetectable during the first 24 h of the life cycle (Fig. S4A). By 27.5 h postinvasion (hpi), Maf1 was detectable in wild-type parasites but was not yet detectable in PB-11 parasites (Fig. S4B). It was detectable, however, in the later hours of the cell cycle at near wild-type levels despite the much lower mRNA abundance. A recent global ribosome profiling study of the Plasmodium falciparum asexual cycle found Maf1 to be translated at very low efficiency from rings through schizonts (38). Perhaps the chimeric 5=UTR generated by the piggyBac insertion removes the cis-acting factors in the native 5=UTR responsible for the observed low translational efficiency, allowing protein production at low mRNA levels, albeit with a notable temporal delay in expression. This endogenous translational control may explain why we were able to obtain stable transfected lines only when expressing Maf1 from a construct designed to reduce protein production levels (Fig. S2).
The piggyBac transposon is well known for its ability to excise cleanly from its insertion site when remobilized (39). We attempted to use this property to restore the wild-type Maf1 locus in the PB-11 mutant line by excising piggyBac. While we were able to obtain parasites stably expressing the piggyBac transposase, no excision of the insertion was detected. Similarly, we attempted to restore the native locus by "crossing out" the piggyBac insertion via CRISPR-Cas9-mediated homologous recombination. We attempted several different strategies with different guide RNAs; however, all failed to restore the wild-type locus.
The PB-11 Maf1 mutant fails to recover from a dormancy-like state. TORC1 is best understood as the central hub from which the progrowth signaling cascades are relayed in response to changes in intracellular levels of amino acids. In yeast, human cells, and Drosophila grown under normal conditions, TORC1 is active and Maf1 is inhibited (20,27,40). When these organisms are starved of amino acids (or TORC1 is inhibited by sirolimus), Maf1 is activated and shuts down Pol III-dependent transcription of tRNAs and other structural RNAs to conserve resources and globally dampen translation. If this safety mechanism is prevented by deletion or silencing of Maf1, death or severe growth defects result upon amino acid starvation.
Over the course of the asexual cell cycle, Plasmodium falciparum resides within its host erythrocyte and imports and digests hemoglobin as its primary source of amino acids. Human hemoglobin lacks isoleucine, and since the parasite does not have the capacity to synthesize it itself, it must import isoleucine from the extracellular environment (i.e., serum in vivo or culture medium in vitro). Isoleucine is the only amino acid needed in the culture medium to sustain growth of P. falciparum in vitro (41).
Unlike the results seen with nutrients such as glucose, whose removal led to rapid death, Babbit et al. found that P. falciparum parasites appear to retard growth and enter a dormancy-like state of low metabolic activity when starved of isoleucine (12). Upon resupplementation, parasite growth resumed seemingly unaffected. Parasites could remain in the absence of isoleucine for upwards of several days and sustain only a minor loss in viability. This is reminiscent of the stationary-phase dormancy observed in yeast, a process for which Maf1 is necessary to maintain viability during prolonged starvation (18,29).
Babbit et al. were unable to identify genetic factors responsible for this phenotype. Neither deletion of the P. falciparum homolog of the GCN2 uncharged tRNA sensor kinase nor the expression of a nonphosphorylatable form of its target, initiation factor 2␣ (IF2␣), had an effect on survival after isoleucine starvation (12). Given the role that Maf1 plays in the amino acid starvation response in other organisms, we investigated how the Maf1 PB-11 mutant responded to isoleucine withdrawal.
Tightly synchronized young ring-stage wild-type and PB-11 mutant parasites were cultured for 72 h in either normal culture medium or medium lacking isoleucine. As previously reported, parasites in the isoleucine-deficient medium stayed at roughly their starting levels of parasitemia whereas those in normal medium grew through 1.5 cell cycles. No detectable difference between wild-type and PB-11 parasites was observed over this period (Fig. 5).
We next tested whether the parasites remained viable over the duration of starvation. Young ring-stage parasites of both wild-type and PB-11 parasites incubated in medium lacking isoleucine for 24 h recovered when returned to normal culture medium for a 72-h recovery period (Fig. 5). However, when the starvation period was increased to 48 h, the PB-11 Maf1 mutant line displayed a notable decrease in recovery, whereas wild-type parasites displayed growth levels similar to those seen with parasites cultured for 72 h in normal medium. With a 72-h starvation, PB-11 mutants displayed no recovery, while wild-type parasites recovered robustly.
Despite not being a full gene disruption, the alteration of Maf1 expression in the PB-11 mutant line appears to prevent the parasites from recovering from prolonged amino acid starvation. To investigate further, we used flow cytometry to track the parasitemia of wild-type and PB-11 cultures every 24 h over 9 days (216 h) of starvation in isoleucine-deficient medium (Fig. 6A). Over the full course of the 9 days, the PB-11 parasitemia decreased at a much higher rate (time required to reach 1/10 the starting level of parasitemia [t 1/10 ] for the WT, 301 h; t 1/10 PB-11, 196 h) (Fig. 6B). Over the first 72 h, however, the death rates of the two lines were nearly identical (t 1/10 WT, 390 h; t 1/10 PB-11, 369 h) (Fig. 6B). As a verification, aliquots of each replicate sample were transferred to normal medium for 3 days of recovery at the 72-h, 144-h, and 216-h points of starvation (Fig. S5). PB-11 displayed no regrowth at any time point, whereas wild-type parasites remained viable and recovered after 9 days.
Babbitt et al. reported that isoleucine starvation in young rings causes cells to progress through the early stages of the cell cycle at approximately 40% the normal rate of growth prior to arrest in the mid-trophozoite stage (12). If PB-11 Maf1 mutant cells can fully recover from a 24-h isoleucine starvation but not at all from a 72-h starvation, this suggests that a specific point in the cell cycle is crossed between those two time points at which the ability to recover is irreversibly compromised, presumably due to altered Maf1 expression.
To map this point, we initiated a culture of early-ring PB-11 parasites in isoleucinedeficient medium for a 144-h incubation (Fig. 6C). Every 8 h, an aliquot of the culture was removed and resuspended in normal medium for a 72-h recovery period. A logistic regression was performed using the relative proportion of parasites recovering at each time point. The results showed a loss of the ability to recover as the duration of the starvation period increased. According to the regression fitted to the data, the point of 50% recovery occurred at 43.1 h of starvation. Given that the cell cycle advances at approximately 40% the normal pace during starvation (12), this corresponds to some time point after 17 h postinvasion. This coincides with the time point at which differences in Maf1 mRNA abundance were most pronounced (Fig. 4) and with the time point just prior to that at which differences in Maf1 protein expression were most pronounced (Fig. S4).
PB-11 displays growth and recovery defects when exposed to multiple growthretarding stressors. In yeast, Maf1 arrests Pol III transcription in response to a broad array of biotic and abiotic stressors in addition to starvation (26). Since TORC1 and, by extension, Maf1 typically regulate progression and arrest of the cell cycle, we investigated whether stressors that stall or retard the cell cycle, in addition to isoleucine starvation, also cause growth or recovery defects for the PB-11 Maf1 mutant.
The antibiotic fosmidomycin inhibits isoprenoid biosynthesis in the Plasmodium apicoplast. Howe et al. previously demonstrated that treatment of parasites with 5 M fosmidomycin causes parasites to seemingly arrest in mid-schizogeny (42). Subsequent treatment of cultures with the downstream isoprenoid geranylgeraniol liberates the parasites from arrest. We arrested wild-type and PB-11 mutant parasites with 5 M fosmidomycin for 72 h and then treated them with 5 M geranylgeraniol for a 72-h regrowth period and compared the levels of recovery to those seen with controls treated with fosmidomycin and geranylgeraniol simultaneously for 72 h (Fig. 7A). The PB-11 Maf1 mutants displayed a pronounced decrease in recovery relative to the wild type.
Ring-stage Plasmodium falciparum is capable of surviving prolonged exposure to temperatures well below 37°C (43). Presumably, the low temperature slows the cell cycle in a manner that may also require proper Maf1 expression for survival. We incubated wild-type and PB-11 Maf1 mutant parasites at 18°C for 72 h and then returned the cultures to 37°C for 72 h of recovery and compared the growth to that seen with controls that remained at 37°C for 72 h (Fig. 7B). Once again, the Maf1 mutant Synchronous young ring-stage PB-11 parasites were washed and transferred to medium lacking isoleucine. Every 3 h, samples were transferred to normal medium for a 72-h recovery period. The final level of parasitemia after recovery was quantified by flow cytometry. A logistic regression fitted to the data shows that each hour of starvation decreases the parasitemia level to 93% of the level seen the previous hour (␤ o ϭ Ϫ0.076, P Ͻ 2.00 ϫ 10 Ϫ16 ). The logistic model fitted to the data predicts that the time point corresponding to 43 h of starvation is the point at which 50% of the PB-11 parasites are able to recover and 50% are not (t 50 ). Data shown are the results of three biological replicates.
PfMaf1 Confers Survival upon Amino Acid Starvation ® parasite line displayed notably decreased recovery from low-temperature stress relative to the wild-type results.
The growth rates of wild-type and Maf1 mutant parasites were distinct, though similar, in culture in normal medium (Fig. 8A). However, cultured in a growth-limiting concentration of isoleucine (8 M [approximately 2% of the normal RPMI 1640 concentration]), the doubling time of the PB-11 Maf1 mutant was markedly longer (Fig. 8B). A similar phenomenon occurred when the parasites were cultured in normal medium  at the stress-inducing temperature of 39°C (Fig. 8C). Interestingly, the growth rates of both the wild-type and PB-11 parasites were slowed by culture in medium containing 20% the normal RPMI 1640 glucose concentration (i.e., 0.4 g/liter), but the conditions used appeared to affect the two parasite clones equally (Fig. 8D). This may suggest that Maf1 is not part of a universal stress response but is instead a component of a specific pathway that responds to certain growth-inhibiting and growth-retarding stressors but not to others.
tRNA expression is dysregulated in the PB-11 Maf1 mutant upon amino acid starvation. Maf1 is the only known regulator of RNA Pol III transcription. When conditions are unfavorable, Maf1 is activated and binds to the Pol III holoenzyme, shutting down transcription of tRNAs, 5S rRNA, and a small number of other structural RNAs. This action is presumed to save cellular resources in times of stress and nutrient limitation and may also globally reduce translational output by limiting the amount of the available tRNA pool. Presumably, Maf1 performs a similar function in Plasmodium falciparum.
The long half-life of mature tRNAs makes it difficult to quantify changes in their expression levels. Typically, pre-tRNAs, immature precursors containing 5= and 3= transcriptional leader sequences and introns, are quantified instead, as these species are rapidly processed, allowing the assessment of Pol III transcriptional activity. At present, no pre-tRNA sequences have been annotated in the Plasmodium falciparum genome. Nearly every eukaryote tyrosine tRNA gene contains an intron immediately 3= of the anticodon, the processing of which is believed to be necessary for the pseudouridine modification found in the anticodon of all mature eukaryote tRNA Tyr molecules (44). We reverse transcribed and cloned the mature Plasmodium falciparum tRNA Tyr gene and found that an 11-nucleotide (nt) sequence adjacent to the anticodon is spliced out in the mature tRNA molecule (Fig. 9A). Using this short intron as a target sequence, we modified a stem-loop real-time quantitative PCR protocol optimized for microRNAs (miRNAs) (45,46) to quantify pre-tRNA abundance using the similarly sized mature 5.8S rRNA (Pol I transcript) as a reference.
Under normal medium conditions, the PB-11 Maf1 mutant parasites displayed higher steady-state expression of pre-tRNA (Fig. 9B), suggesting that Maf1 may play a role in regulating tRNA expression in normal growth as well as under conditions of stress. After 24 h of isoleucine deprivation, the difference in pre-tRNA levels was even more pronounced (Fig. 9B), supporting the notion that the failure of Maf1-dependent regulation of tRNA expression contributes to this mutant's inability to recover from amino acid starvation and other forms of stress.
It has been suggested that the increase in intracellular tRNA levels caused by genetic suppression of Maf1 can cause a global increase in translation due to increased levels of the initiator tRNA (47). Regulating translation in times of stress is a fundamental component of most cellular stress responses because it allows cells to conserve resources, decrease the buildup of aggregates and unfolded proteins, and selectively translate specific subsets of mRNAs (48). If this process is counteracted by excessive translational initiation, the cell's ability to survive stress may be seriously compromised.
To quantify global translation activity, we used a puromycylation assay (49). Briefly, the translation inhibitor puromycin incorporates into nascent peptide chains on actively translating ribosomes. A monoclonal antibody specific for puromycin can then be used to quantify levels of puromycin incorporation as a surrogate for global translational output. We pulsed parasite cultures for 1 h with puromycin prior to harvesting protein and then quantified puromycin incorporation by enzyme-linked immunosorbent assay (ELISA) using anti-P. falciparum glyceraldehyde-3-phosphate dehydrogenase genes (anti-PfGAPDH) as an internal control.
When cells were cultured in normal medium, the PB-11 Maf1 mutant displayed moderately higher puromycin incorporation than the wild-type cells (Fig. 9C). However, after 24 h of isoleucine starvation, the PB-11 mutant displayed a dramatic increase in puromycin incorporation relative to the wild-type control. This inability to repress translation under conditions of amino acid starvation may further contribute to the inability of the mutant line to survive prolonged starvation.
An artemisinin-resistant K13 variant confers increased survival and greater control of pre-tRNA expression upon isoleucine starvation. Artemisinin and its derivatives induce growth retardation in ring-stage parasites (9), and artemisininresistant isolates appear capable of recovering from this drug-induced growth inhibition whereas sensitive isolates cannot (10). We investigated whether this ability to overcome growth retardation translated to increased recovery from prolonged amino acid starvation and whether drug-resistant parasites displayed differential levels of regulation of Pol III transcription as a potential point of connection between the artemisinin resistance phenotype and PfMaf1.
Using an artemisinin-resistant field isolate originating from Battambang, Cambodia (K13 allele: R539T), and a syngeneic line in which the K13 was reverted to the sensitive allele (50), we cultured the parasites in the absence of isoleucine and evaluated their ability to regulate pre-tRNA expression and recover from starvation (Fig. 10A). After 24 h of isoleucine starvation, the R539T-resistant isolate displayed 27% the expression level of pre-tRNA seen with the syngeneic revertant line. Consistent with our observation in the PB-11 Maf1 mutant, the ability to suppress pre-tRNA expression coincided with increased survival. After 72 h of isoleucine starvation, and a subsequent 72 h of recovery, the R539T-resistant isolate displayed recovery at a level nearly 3-fold greater than that seen with its revertant counterpart (Fig. 10B).
We additionally assessed whether the PB-11 Maf1 mutant, defective in pre-tRNA regulation, displays any difference from the NF54 wild-type control with respect to and Maf1 mutant (PB-11) parasites. Synchronous young ring-stage parasites were incubated in normal medium or medium lacking isoleucine for 24 h prior to RNA isolation. Pre-tRNA Tyr expression was quantified relative to 5.8S rRNA levels. (C) ELISA of puromycin incorporation relative to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene levels in wild-type and mutant parasites under conditions of the indicated treatments. Synchronous young ring-stage parasites were incubated in normal medium or medium lacking isoleucine for 24 h prior to the 1-h puromycin pulse and subsequent harvesting. Translational arrest by cycloheximide (CHX) treatment was used as a negative control. P values were calculated using t tests of three biological replicates.
McLean and Jacobs-Lorena ® recovery from artemisinin treatment. Using a 50% effective concentration (EC 50 ) assay derived from the ring-stage survival assay (51), the PB-11 Maf1 mutant parasite line displayed poorer recovery in the presence of dihydroartemisinin than the wild-type NF54 parasites (Fig. S6). Together, these results suggest there may be overlap between the pathways that regulate growth and recovery from both artemisinin treatment and amino acid starvation.

DISCUSSION
The TORC1 pathway is a highly conserved signaling pathway that allows eukaryotes to modulate cellular growth in response to changes in the environment. Most of the recognizable components of this pathway have been lost in the Plasmodium genus. In this study, we showed that the P. falciparum ortholog of Maf1 contributes to survival and recovery of parasites under conditions of various forms of growth-inhibiting stress. The observed phenotypes are consistent with those reported for other organisms, suggesting that, despite the loss of the TORC1 pathway, Plasmodium falciparum encodes an analogous growth regulation pathway that acts upon this highly conserved repressor of Pol III transcription.
To study Maf1, we used a transposon insertion mutant produced as part of an effort to generate a genome-wide mutant collection (34). While this mutant was not a genetic knockout as previously reported, insertion of the piggyBac element into the Maf1 5=UTR resulted in profound changes in both mRNA and protein expression. This observation should serve as a reminder that the regulatory elements within transposons can affect nearby loci and that phenotypes observed after transposon insertion may be due to factors other than insertional mutagenesis alone. It remains a formal possibility that the phenotype reported here for the PB-11 mutant may have been caused in part by secondary effects on other loci not studied here. However, our observation of the mutant's inability to regulate pre-tRNA expression and protein synthesis, phenotypes previously described for yeast, human, and Drosophila Maf1 mutants (20,27,40), supports the conclusion that the mutant's altered Maf1 expression profile is responsible for its inability to recover from isoleucine starvation and other forms of stress.
Despite intensive efforts, we were unable to generate a knockout parasite line using standard methods, suggesting that basal Maf1 expression may be essential for in vitro asexual growth. The reason that PfMaf1 would be necessary for growth under normal conditions is not clear, as Maf1 knockouts are viable in both yeast and mice (24, 52). One possible explanation may stem from the observation that the PB-11 Maf1 mutant PfMaf1 Confers Survival upon Amino Acid Starvation ® displayed elevated levels of global translation relative to the wild-type control even under normal growth conditions. Plasmodium falciparum schizonts are reported to globally suppress translation at the late stages of the intraerythrocytic cycle (53). It is possible that some degree of PfMaf1 suppression of tRNA and 5S rRNA transcription is needed for the maintenance of this suppression.
In yeast and human cells, the association of Maf1 with the RNA Pol III complex is regulated by phosphorylation. Determining whether PfMaf1 is also phosphoregulated and identifying the kinases and phosphatases responsible for its regulation will be important next steps in understanding parasite growth regulation.
The description by Babbitt et al. of the dormancy-like slow growth response of P. falciparum upon isoleucine starvation was the first thorough documentation of the parasite's ability to modulate its growth rate in response to environmental changes (12). However, after being unable to implicate any specific molecular pathways in the response (i.e., GCN2/PfIK2-IF2␣), and after observing a dose-dependent correlation between the isoleucine concentration and the growth rate, they concluded that the parasite does not actively regulate its growth and that isoleucine is likely the ratelimiting factor in the progression of the cell cycle. We feel that this conclusion might have been premature. Similar conclusions were drawn in studies of other organisms, such as yeast, until it was demonstrated that chemical or genetic inhibition of TOR induced the same transition into stationary phase as nutrient starvation and that partial inhibition of TOR slowed the cell cycle in a manner similar to that seen with nutrient limitation (13,17,18). We believe that our description of the role of PfMaf1 in maintaining viability upon growth retardation is the first evidence for the existence of an analogous pathway in P. falciparum. The remaining components of the pathway, however, have yet to be identified.
Unlike TOR, Maf1 has never been shown to arrest or retard growth on its own. It is instead needed to maintain viability during the period of stress or nutrient limitation, much as we observed in P. falciparum. It is intriguing that the strain with the single point mutation in the K13 gene that confers artemisinin survival also displayed increased survival upon prolonged amino acid starvation as well as greater suppression of tRNA expression. If growth is actively regulated in Plasmodium by a yet-to-be determined pathway, then the effects of the K13 mutation may allow the parasite to shut down growth more effectively when faced with undesirable circumstances or to resume growth more effectively after a growth-retarding insult.

MATERIALS AND METHODS
Parasites. The NF54 attB (used as the "wild-type" strain in this study) (54) and Dd2 attB (30) strains were generously provided from the laboratory of David Fidock (Columbia University). The K13 R539T artemisinin-resistant isolate (Cam3.I IPC 5202) was isolated from a patient in Battambang Province, Cambodia, in 2011. It was obtained from the Malaria Research and Reference Reagent Resource Center (MR4) for distribution by BEI Resources NIAID, NIH (product number MRA-1240), and was originally contributed by Didier Ménard (Institut Pasteur). The corresponding revertant line, Cam3.I_rev, was also obtained from BEI Resources NIAID, NIH (product number MRA-1252), and was contributed by David A. Fidock (Columbia University) (50). The PB-11 Maf1 mutant parasite line was obtained by special request from BEI Resources NIAID, NIH (product number MRA-1031), and was contributed by John Adams (34).
Yeast complementation. A full-length Plasmodium falciparum Maf1 coding sequence was synthesized with codon usage optimized for S. cerevisiae expression (GenScript). Using Gibson assembly (NEB; catalog no. E2611S), the core region of PfMaf1 (amino acids 237 to 346) was fused between the S. cerevisiae N terminus (amino acids 1 to 233) and C-terminal "acidic tail" (amino acids 338 to 395). The resulting chimeric open reading frame was cloned into the BamHI and XhoI sites of the p416-Met25-3xHA yeast expression vector. The full-length yeast Maf1 open reading frame (YDR005C) was similarly cloned into the same vector. The Maf1Δ knockout line (BY4741 MATa library; ΔYDR005C) was transformed using the LiAc/ssDNA/PEG method (55). Survival assays were carried out on synthetic minimal dextrose medium lacking uracil (for plasmid selection) and methionine (to derepress the Met25 promoter for transgene expression) with or without 10 nM sirolimus (Cell Signaling Technologies; catalog no. 9904).
Flow cytometry. For each sample to be analyzed, 1 ml of parasite culture was pelleted and resuspended in 1 ml of phosphate-buffered saline (PBS) containing 4% (wt/vol) paraformaldehyde. Samples were rocked at 4°C for 20 to 30 h of fixation. Samples were then pelleted and resuspended in PBS containing 0.1% (vol/vol) Triton X-100 and rocked at room temperature for 1 h for permeabilization. This process was then repeated three additional times with PBS (without Triton X-100) to remove as much hemoglobin as possible. Samples were then diluted approximately 100-fold into normal culture medium containing 1ϫ SYBR green I (Thermo Fisher Scientific; catalog no. S7563) and analyzed on a FACSCalibur (BD Biosciences) flow cytometer. The resulting data were further analyzed using FlowJo 10.0.7 analysis software.
Transfection. The DD24-Maf1 construct was synthesized (GenScript) with an N-terminal 3ϫ hemagglutinin (HA) tag followed by a DD24 destabilization domain and a yeast codon-optimized Maf1 open reading frame (the same synthesized construct used for yeast complementation). This open reading frame was cloned into the XhoI and XmaI sites of vector pLN-mRPL2pr to generate pLN-mRPL2pr-DD24-Maf1. This vector was used to transfect Dd2 attB parasites (30) by the pre-loading method (57) using a modified Bxb1 integrase transfection method (58). Stably transfected cells were maintained with 2.5 g/ml blasticidin S. The Shield ligand was obtained from Aobious (catalog no. 1848).
Genome sequencing. High-quality, high-molecular-mass genomic DNA was prepared from 400 ml of PB-11 culture (2% hematocrit,~10% parasitemia) enriched in late-stage parasites. Parasites were liberated from erythrocytes by saponin treatment and washed twice with PBS to remove excess hemoglobin. The resulting parasite pellet was used as starting material for genomic DNA isolation using a Blood and Cell Culture minikit (Qiagen; catalog no. 13323). Genomic DNA was sent to Axeq Technologies (Seoul, South Korea) care of the Macrogen Corporation (MD, USA) for library preparation and sequencing. An amplification-free library (59) was sequenced by paired-end sequencing to 73ϫ coverage using an Illumina HiSeq 2000 sequencing system.
Sequencing analysis. The analysis of the sequencing data largely followed the protocol described by Balu et al. (35), who had previously sequenced the NF54 clone that was used as the parental line in the transposon screen which produced the PB-11 mutant (34). Two sets of paired-end reads from the parental NF54 clone (ERS038926 and ERS184445), produced by two different sequencing platforms, were obtained from the European Nucleotide Archive. These reads were used to "update" the PlasmoDB-26 P. falciparum 3D7 genome sequence using ICORN as part of the PAGIT software package (60). Seven iterations of ICORN generated 561 1-bp substitutions and 774 small insertions or deletions across the entire reference genome. The PB-11 reads were aligned to this updated reference sequence using bwa (61) with the default parameters. Approximately 82% of the reads aligned to the reference sequence. The resulting sam file was sorted, indexed, and filtered of duplicates using Picard-tools v 1.119. Reads were realigned around indels using the Genome Analysis Toolkit (GATK) (62), and raw variants were called using the GATK HaploTypeCaller, with ploidy set to 1. The raw variants were then filtered using vcftools v0. 1.13 (63) to include only calls with a quality score greater than 60 and a minimum depth of 10 reads. The program snpEff v3 (64) was then used to filter only those variants found within open reading frames. This resulted in 32 potential variants within open reading frames. However, upon visual inspection using the Integrated Genomics Viewer v2.3.9 (65), every call was found in either a low-complexity region (e.g., an extended mononucleotide tract) or a repetitive region. In each case, there was read support for the reference sequence, leaving us to conclude that these variants more than likely represented false positives. To verify that our pipeline was indeed capable of detecting variants, we reran the pipeline analysis using the reads from the C9 piggyBac insertion mutant sequenced by Balu et al. (ENA ERS038913) (35). We were able to easily detect the same two SNPs reported previously. mRNA expression qPCR time course. Synchronous cultures were established by sequential sorbitol treatments. For each time point, 50 ml of 2% hematocrit culture at approximately 10% parasitemia was pelleted, immediately disrupted with 10 ml of TRIzol (Thermo Fisher Scientific; catalog no. 15596018), and used for subsequent phenol-chloroform extraction and ethanol precipitation per the manufacturer's protocol. First-strand cDNA synthesis was performed using a mixture of oligo(dT) and random hexamers and Superscript III (Thermo Fisher Scientific; catalog no. 18080044) and the standard protocol. For quantitative PCR, the Maf1 transcript was amplified using forward primer 5= GATGCCCACGATCGTTT TAT 3= and reverse primer 5= CGGAGCTAAATATTTGTGTATTGC 3=. The seryl-tRNA ligase transcript, PF3D7_0717700, was used as an internal control, with 5= AAGTAGCAGGTCATCGTGGTT 3= as a forward primer and 5= TTCGGCACATTCTTCCATAA 3= as a reverse primer. Quantitative PCR was performed using SYBR green PCR master mix (Thermo Fisher Scientific; catalog no. 4309155) on a StepOnePlus (Thermo Fisher Scientific) real-time PCR system. 5=-RACE. RNA was extracted from 50 ml of a 2% hematocrit-10% parasitemia mixed-stage culture using TRIzol (Thermo Fisher Scientific; catalog no. 15596018). 5=-RACE was performed using a FirstChoice RLM-RACE kit (Thermo Fisher Scientific; catalog no. AM1700). The resulting PCR product was cloned using a CloneJET PCR cloning kit (Thermo Fisher Scientific; catalog no. K1231) and sequenced (Macrogen Corporation, USA) to determine the 5= transcription start site.

Recombinant protein expression and antibody production.
For recombinant protein production, the Plasmodium berghei Maf1 ortholog (PBANKA_0718500) was selected instead of the P. falciparum ortholog, as it is much smaller and contains a short repeat in place of the P. falciparum asparagine-rich region. Outside this region, the sequences are nearly identical. The P. berghei Maf1 sequence was synthesized with Escherichia coli-optimized codon usage (DNA2.0) with an N-terminal 6ϫHis tag in a T7 promoter expression plasmid. Protein expression was conducted in NiCo21(DE3) cells (NEB; catalog no. C2529). Expression was performed using 1-liter cultures induced at an optical density at 600 nm (OD 600 ) of 3.0 with 0.5 mM IPTG (isopropyl-␤-D-thiogalactopyranoside)-Terrific broth for 20 h at room temperature. Cells were lysed by sonication, and protein was isolated using nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen; catalog no. 30210). The recombinant protein was used to produce antiserum from two separate guinea pigs by Cocalico Biologicals (PA, USA) using the standard protocol. Antiserum was affinity isolated against the recombinant PbMaf1 protein. Western blots were visualized using a Li-COR Odyssey imaging system with 680RD linked anti-mouse (Li-COR; catalog no. 925-68070) and anti-guinea pig (Li-COR; catalog no. 925-68077) secondary antibodies. The 1-10 b PfGAPDH monoclonal (mouse) (66) antibody was used as a loading control.
Growth curves. Asynchronous parasite cultures were maintained under the conditions described and sampled every 48 h to quantify parasitemia by flow cytometry. Cultures were then diluted between 2-fold and 10-fold to maintain a parasitemia level of between 0.5% and 1.0%, and the dilution factor was recorded. The growth at each time point was measured as the parasitemia level multiplied by the dilution factor of the previous dilution. The final data set was normalized such that the initial level of parasitemia for each line was given a value of 1.0 to account for differences in the starting levels of parasitemia of the different parasite lines. A two-factor regression of the approximate hour (ln[adjusted parasitemia]) plus the genotype was fitted for each condition to test the effect of the parasite genotype on the growth rate, with time as a covariate. Doubling time (t D ) was calculated for each line from single-factor regressions against time alone, using t D ϭ log 2/␤ o , where ␤ o is the coefficient from the single-factor regression.
tRNA Tyr intron verification and stem-loop PCR. For tRNA analysis, RNA was isolated using a miRvana miRNA isolation kit (Thermo Fisher Scientific; catalog no. AM1560) to specifically enrich small RNAs. Due to the inherently high stability of the tRNA secondary structure, a modified reversetranscription protocol was employed. The total volume of small RNA was mixed with a 5 M concentration of RT primer 5= TTACTTGTACAGCTCGTCCATGCCGAGATCCGATGAACCGGAATCG 3=, which anneals to the 3= end of the tRNA Tyr (PF3D7_0702800) sequence but contains a RACE-like 5= extension of an unrelated primer sequence for subsequent PCR amplification (to avoid amplifying the genomic locus). The small-RNA-primer mixture was heated to 98°C for 5 min and then cooled slowly to 4°C. Reverse transcription was then performed using Superscript III (Thermo Fisher Scientific; catalog no. 18080044) and the manufacturer's protocol, with the exception that it was performed at the elevated temperature of 65°C for 1 h to minimize the effects of the presence of the tRNA secondary structure. The cDNA was then amplified using Phusion High-Fidelity polymerase (NEB; catalog no. M0530L), forward primer 5= CCGATGATAGCTCAGTTGGTAGA 3= (which anneals to the 5= end of the tRNA Tyr sequence), and reverse primer 5= CTTGTACAGCTCGTCCATGCC 3= (which anneals to the 5= extension of the RT primer). The resulting PCR product was cloned using a CloneJET PCR Cloning kit (Thermo Fisher Scientific; catalog no. K1231) and sequenced (Macrogen Corporation) to verify the excision of the predicted 11-nt intron.
Pre-tRNA Tyr stem-loop qPCR. For analysis of pre-tRNA Tyr , highly synchronous parasite cultures were initiated by Percoll-sorbitol treatment to generate 50 ml of 2% hematocrit culture with a level of parasitemia of between 5% and 10% and with a maximum parasite age of 4 hpi. These cultures were then washed three times in 50 ml PBS and resuspended in either normal media or media lacking isoleucine and then incubated for 24 h prior to harvesting. Cultures were pelleted, treated with 0.1% saponin-PBS, and washed twice with 50 ml PBS to remove liberated hemoglobin. A small RNA sample was then isolated from parasite pellets using a miRvana miRNA isolation kit (Thermo Fisher Scientific; catalog no. AM1560) according to the manufacturer's protocol. In order to specifically detect the presence of the 11-nt intron in the pre-tRNA Tyr , a stem-loop PCR method originally developed for miRNA detection (45,46) was adapted. In this method, a self-annealing 5= stem-loop extension to the RT primer adds length to the final product for downstream qPCR amplification product and may also provide additional stability for the annealing of the short complementary primer sequence due to stacking interactions provided by the stem-loop. For reverse transcription, small RNA samples were combined with a 5 M mixture of the stem-loop RT primers 5= GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACATAACCATTTC 3= and GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCAGCATTCTGA, which anneal to the pre-tRNA Tyr intron and the 3= end of 5.8S rRNA, respectively. The primer-RNA mixture was heated to 98°C for 5 min and then cooled to 4°C. Reverse transcription was carried out at 65°C for 1 h to process the RNA secondary structure correctly but otherwise followed the Superscript III protocol of the manufacturer (Thermo Fisher Scientific; catalog no. 18080044). Quantitative PCR was performed using the common reverse primer 5= CAGTGCAGGGTCCGAGGT 3= (which anneals to the shared stem-loop structure) and the gene-specific forward primers 5= AGTTGGTAGAGCGGCAGACT 3= (pre-tRNA Tyr ) and 5= AGCAAAACGCGA TAAGCAAT 3= (5.8S rRNA) on a StepOnePlus real-time PCR system (Thermo Fisher Scientific) and Power SYBR green master mix (Thermo Fisher Scientific; catalog no. 4367659).
Puromycylation assay. Cultures were initiated using Percoll-sorbitol synchronization to produce 50-ml cultures at 2% hematocrit and approximately 10% parasitemia. Cultures were then pelleted and washed three times with 50 ml PBS and resuspended in either normal or isoleucine-lacking medium. After 24 h, puromycin was added to reach a final concentration of 1.0 M and parasites were incubated for an additional 1 h. Samples were then pelleted and resuspended in 50 ml 0.1% saponin-PBS and were subsequently washed twice with 50 ml PBS to remove hemoglobin. Parasite pellets were then lysed using radioimmunoprecipitation assay (RIPA) buffer containing protease inhibitors. For ELISA, 500 ng/well of each biological replicate was separately probed with either the anti-puromycin monoclonal antibody 3RH11 (Kerafast; catalog no. EQ0001) or monoclonal anti-Plasmodium falciparum GAPDH as an internal control. As a negative control, parasites grown in normal media were treated with 10 g/ml cycloheximide simulataneously with puromycin to arrest protein synthesis and block puromycin incorporation.
Ring-stage-survival EC 50 assay. Schizonts were isolated from Percoll gradients and allowed to invade fresh erythrocytes for 3 h, after which a sorbitol synchronization was performed. Synchronous 0-to-3-h-old rings were immediately divided into 1-ml (2% hematocrit) volumes of replicate wells and treated with 12 steps of dihydroartemisinin diluted 3-fold in dimethyl sulfoxide (DMSO) beginning with a maximum concentration of 10 M. After 6 h of drug exposure, the contents of each 1-ml well of culture were pelleted and washed with 15 ml of fresh medium to remove drug and then returned to a 1-ml culture for an additional 90 h of recovery. At the end of the 96-h assay, cultures were fixed and quantified by flow cytometry as described above. EC 50 values were calculated using a four-parameter log-logistic function and the drc package in R (67).
Statistical analysis. All statistical analyses were performed using R version 3.2.3 (10 December 2015) (68). t tests, linear and logistic regressions, and some figures were generated using R base functions. The drc package (67) was used for determining the 50% survival point of PB-11, and the ggplot2 package (69) was used for the production of several figures.
Accession number(s). The paired-end read data determined in this work have been deposited in the NIH Sequence Read Archive under accession number SRR4206195.