Purine-responsive expression of the Leishmania donovani NT3 purine nucleobase transporter is mediated by a conserved RNA stem-loop

The ability to modulate gene expression in response to changes in the host environment is essential for survival of the kinetoplastid parasite Leishmania. Unlike most eukaryotes, gene expression in kinetoplastids is predominately regulated post-transcriptionally. Consequently, RNA-binding proteins (RBPs) and mRNA-encoded sequence elements serve as primary determinants of gene regulation in these organisms; however, few have been ascribed roles in specific stress-response pathways. Leishmania lack the capacity for de novo purine synthesis and must scavenge these essential nutrients from the host. Leishmania have evolved a robust stress response to withstand sustained periods of purine scarcity during their lifecycle. The purine nucleobase transporter, LdNT3, is among the most substantially upregulated proteins in purine-starved L. donovani. Here we report that the post-translational stability of the LdNT3 protein is unchanged in response to purine starvation. Instead, LdNT3 upregulation is primarily mediated by a 33 nucleotide (nt) sequence in the LdNT3 mRNA 3’-untranslated region that is predicted to adopt a stem-loop structure. While this sequence is highly conserved within the mRNAs of orthologous transporters in multiple kinetoplastid species, putative stem-loops from L. donovani and Trypanosoma brucei nucleobase transporter mRNAs are not functionally interchangeable for purine-responsive regulation. Through mutational analysis of the element, we demonstrate that species specificity is attributable to just three variant bases within the predicted loop. Finally, we provide evidence that the abundance of the trans-acting factor that binds the LdNT3 stem-loop in vivo is substantially higher than required for regulation of LdNT3 alone, implying a potential role in regulating other purine-responsive genes.

Kinetoplastid parasites of the genus Leishmania are the etiological agents of leishmaniasis, a suite of debilitating and often fatal diseases that affect roughly 12 million people worldwide (1). Over the course of their lifecycles, Leishmania exist as both extracellular promastigotes in the ambient midgut of a sandfly vector and as intracellular amastigotes in the phagolysosomes of mammalian macrophages. As these compartments differ substantially in pH, temperature, and nutrient availability, Leishmania must undergo dramatic metabolic and physiological transformations to adapt to life in their respective hosts (2,3). Such environmental responses are so intertwined with the leishmanial lifecycle that several stressors function as cues for parasite differentiation, triggering the transition from one developmental stage to the next (4)(5)(6)(7)(8). Despite their importance to parasite biology, relatively little is known of the molecular mechanisms that allow Leishmania to respond to fluctuations in the extracellular milieu.
Leishmania and other kinetoplastid protozoa separated from the eukaryotic lineage early in evolutionary history and consequently exhibit several unique biological features, including their means of gene expression. Deviating from the onepromotor-to-one-gene canon of higher eukaryotes, the genomes of kinetoplastid parasites are organized as large unidirectional transcription units that can comprise tens to hundreds of tandemly arranged genes and span up to 100 kb in length (9). Transcription by RNA polymerase II results in the production of polycistronic primary transcripts, which mature into single-gene mRNAs through 3' polyadenylation and trans-splicing of a capped exon (the spliced leader) to the 5' end of each coding sequence (CDS) (reviewed in 10). As a result, Leishmania rely almost exclusively on posttranscriptional control points like mRNA stability and localization, translational efficiency, and protein half-life to respond to exogenous stimuli. As in higher eukaryotes, transcript abundance and translation in the kinetoplastids are modulated by the interactions of trans-acting RBPs and cis-acting regulatory elements encoded in mRNA UTRs or the CDS itself. In recent decades, much effort has been devoted to the study of such elements in these organisms; however, only a handful of RBPs and even fewer discrete mRNA elements have been identified as both necessary and sufficient for gene regulation in the context of specific stress response pathways (11)(12)(13).
Leishmania are unable to synthesize purines de novo and must scavenge these essential nutrients from the host (14,15). The availability of salvageable purines likely fluctuates throughout the parasite lifecycle and Leishmania have evolved to withstand periods of extreme purine scarcity. Indeed, this particular type of nutrient stress may be an important regulator of parasite infectivity. Adenosine supplementation negatively impacts the efficiency of in vivo metacyclogenesis, the process wherein poorly infective procyclic promastigotes in the sandfly midgut differentiate into highly infective metacyclic promastigotes, primed for survival within a vertebrate host (8). A robust and reproducible purine stress response is easily induced in vitro by omission of purines from the growth medium, and we showed that purine-starved Leishmania donovani promastigotes enter a reversible quiescent-like state in which they can persist for over three months (14,16,17). Whole proteome comparisons of purine-starved and -replete L. donovani promastigotes conducted over a 6-to 48-hour window revealed a temporallycontrolled remodeling of the cellular proteome. While early changes centered on increasing purine uptake, later responses reflected a larger restructuring of cellular metabolism to reduce energy expenditure and enhance general stress tolerance. The global transcriptomes of these cells were also significantly different; however, changes in mRNA abundance often tracked poorly with those manifested at the protein level, implicating both translational and post-translational mechanisms (16). Together, these analyses demonstrated that purine-responsive cellular remodeling is complex and orchestrated at multiple levels of post-transcriptional regulation.
Though our earlier global analyses provided critical insight into what processes are important for adaptation to purine starvation, they uncovered little about how specific changes in protein abundance are established. In this report, we look deeper at the molecular mechanisms underlying the leishmanial purine stress response through the lens of a representative purine-responsive gene. The L. donovani purine nucleobase transporter 3 (LdNT3) is one of the earliest and most substantially upregulated proteins in purine-starved L. donovani. LdNT3 upregulation is mediated in part at the levels of mRNA abundance and translational efficiency (16,18). Here we demonstrate that LdNT3 protein stability is not altered under purine stress, establishing the dominance of mRNA-level and translational control points in purine-responsive LdNT3 regulation. We identify a 33 nt predicted stem-loop sequence in the LdNT3 3'-UTR (referred to as the LdNT3 stem-loop) that represses expression when extracellular purines are abundant. Using a series of integrating luciferase reporter constructs, we show that the LdNT3 stemloop is sufficient to confer purine-responsive regulation in heterologous sequence contexts. We examine evolutionary conservation of the element and conduct a thorough mutational analysis to identify functionally important regions required for repressor activity in purine-replete L. donovani. Lastly, we show that the LdNT3 stem-loop is sufficient to confer purine-responsiveness to a highabundance transcript, suggesting that the cognate RBP responsible for binding this element in vivo is present within the cell in substantial excess of what is required to regulate LdNT3 expression alone.

RESULTS
LdNT3 protein stability is not regulated in response to purine stress. To elucidate the molecular mechanisms that coordinate purineresponsive LdNT3 expression, we first determined the levels at which they operate. It is published that changes in both mRNA stability and translational efficiency contribute to LdNT3 upregulation under purine starvation (16,18). However, our earlier studies ignored the potential contribution of posttranslational stabilization. In our experience, epitope-tagged LdNT3 is refractory to direct detection by western blotting at the low levels expressed in purine-replete cultures. Therefore, we modified a dual-luciferase system established in our laboratory to indirectly measure changes in LdNT3 protein stability via enzymatic reporter assay.
In the simplest iteration of the dual-luciferase system (described at length in 18), the firefly luciferase gene (Fluc) is fused in-frame with a selectable drug resistance marker and integrated in place of one allelic copy of the gene of interest. Reporter integration fully replaces the CDS of the targeted gene while preserving the endogenous intergenic regions (IGRs), which contain the requisite signals for trans-splicing and polyadenylation. Importantly, as the 5' and 3' mRNA UTRs are derived from these up-and downstream IGRs, respectively, their contributions to regulation are reflected in luciferase activity. To normalize Fluc activity between replicates, Renilla luciferase (Rluc) is similarly integrated into the locus of a gene for which expression does not change under the conditions of the experiment. In probing the purine stress response, we used UMP synthase (UMPS) as an unresponsive control, since neither UMPS mRNA nor protein abundance are affected by purine starvation (16,18). To adapt this system for the study of posttranslational stability, we generated cell lines in which the LdNT3 CDS was fused via its N-terminus to a Fluc reporter and integrated into either its endogenous locus or that of a purine-unresponsive control, LdNT4 ( Figure 1A). The multicistronic constructs used for integration contained a blasticidin resistance gene (BSD) to facilitate mutant selection and a 2A peptide from the Thosea asigna virus (2A) that is co-translationally cleaved, liberating the BSD-2A polypeptide to minimize the size of appended tag on LdNT3 (19,20). In this configuration, the post-translational fate of the reporter is coupled to that of the transporter while mRNA stability and translation are governed by the native LdNT3 UTRs and/or CDS such that changes in luciferase activity reflect the cumulative effect of mechanisms operating at all post-transcriptional levels. In contrast, the purine nucleobase transporter 4 (LdNT4), while homologous to LdNT3, is not differentially regulated with respect to purine availability and Fluc expression from this locus consistently reflects an absence of purine sensitivity (16). Thus, expression of Fluc-LdNT3 integrated into this locus reflects only regulation conferred by elements contained within the LdNT3 CDS itself or directly affecting stability of the protein. As anticipated, endogenously tagged Fluc-LdNT3 was significantly upregulated by 24 hours of purine starvation, consistent with previous experiments that implicated the UTRs in LdNT3 regulation. In cells expressing Fluc-LdNT3 flanked by neutral LdNT4 UTRs, however, luciferase activity was not affected by purine stress ( Figure  1B). These data indicate that the LdNT3 CDS does not encode additional cis-acting purine response elements nor does protein stability contribute to LdNT3 upregulation in purine-starved parasites. Thus, in the absence of post-translational control, transcript stability and translational efficiency serve as primary control points mediating purineresponsive changes in LdNT3 abundance.
A 33 nt stem loop in the LdNT3 mRNA 3'-UTR represses expression under purine-replete conditions. The LdNT3 5'-and 3'-UTRs are together sufficient to confer purine-responsiveness to a reporter (16,18), strongly implicating the presence of cis-acting regulatory sites within one or both of these regions. The 3'-UTR of the orthologous Trypanosoma brucei purine nucleobase transporter, TbNT8.1, encodes a predicted stem-loop that both is necessary and sufficient to repress TbNT8.1 expression when extracellular purines are abundant (13). Based on homology to this region, we identified a similar 33 nt predicted stem-loop in the LdNT3 3'-UTR, approximately 2.73 kb downstream of the stop codon. (Figures 2A and 2B). While the exact secondary structure of this element in vivo has not been formally demonstrated, we will refer to these sequences as stem-loops throughout for convenience. Though absent from the UTRs of other purine-responsive genes in L. donovani 1 , this element was identified in the UTRs of orthologous purine transporters from a variety of kinetoplastids, suggesting a strong evolutionary pressure for conservation ( Figure 2A; Figure S1). To test whether this region also confers purineresponsiveness in L. donovani, we generated cell lines in which a Fluc-BSD transgene was expressed from the endogenous LdNT3 locus under the control of either wildtype UTRs or a modified 3'-UTR lacking the putative stem-loop ( Figure 2B). Parasites expressing Fluc-BSD flanked by wildtype UTRs demonstrated an approximate 12-fold increase in luciferase activity in response to purine stress. The magnitude of this effect was diminished by nearly ninety percent in stem-loop deletion mutants, wherein Fluc activity was increased only ~1.4-fold by starvation. Specifically, deletion of the stem-loop resulted in a ~11-fold increase in basal Fluc expression under purine-replete conditions, consistent with the sequence functioning as a negative regulator ( Figure 2C). Together, these data implicate the repressive LdNT3 stem-loop is a major regulator of purine-responsive LdNT3 expression.
We next asked whether this RNA element was sufficient to confer regulation to an Fluc-BSD reporter expressed from the LdNT4 locus, which is normally not affected by purine stress. We integrated constructs encoding a Fluc-BSD reporter flanked by either wildtype LdNT4 UTRs or a 3'-UTR harboring the LdNT3 stem-loop at one of two different positions ( Figure 3A) that were predicted via Mfold to preserve stem-loop folding (21). As shown in Figure 3B, the effects of stem-loop insertion were position-dependent. Placing the stem-loop 242 bases into the LdNT4 3'-UTR led to a significant decrease in basal Fluc-BSD expression, which translated to a ~6-fold increase in luciferase activity under purine-restricted conditions. In contrast, when inserted at position +419 the stem-loop did not confer differential expression, possibly reflecting an inability of the element to fold properly in this genetic context. Thus, the LdNT3 stem-loop is sufficient for purineresponsive expression but is sensitive to sequence context. 1 Data not shown.

Regulation by the stem-loop is species-specific and depends upon conserved residues in the loop.
The purine-response elements from TbNT8.1 and LdNT3 share a 33 nt core with 80% identity (Figures 2A and 4A). To determine if the orthologous TbNT8.1 stem-loop was functional in L. donovani, we modified the reporter construct depicted in Figure 3A to insert the minimal TbNT8.1 stem-loop at position +242 in the LdNT4 3'-UTR. While this element was sufficient for regulation in T. brucei (16), it was unable to confer purine-responsive expression to the Fluc-BSD reporter in L. donovani ( Figure 4B), suggesting that the RBPs that associate with these elements in L. donovani and T. brucei have different binding specificities.
To determine if the inability of the TbNT8.1 element to function in Leishmania is due to differences in the sequence of the stem, loop, or both regions, we generated chimeras in which the TbNT8.1 loop sequence was appended to the stem of the LdNT3 ortholog, and vice versa. These chimeric stem-loops were inserted at position +242 in the LdNT4 Fluc-BSD constructs and integrated into the LdNT4 locus of a dual-luciferase compatible cell line. Parasites encoding a TbNT8.1 loop on an LdNT3 stem demonstrated a partial reduction in luciferase activity under purine replete conditions compared to the TbNT8.1 stem-loop control, but there was no increase in expression upon purine starvation ( Figure 4C). In contrast, Fluc-BSD expression was significantly repressed by the reciprocal LdNT3 loop-TbNT8.1 stem mutant in the presence of exogenous purine. This effect was reversed by 24 hours of purine stress, resulting in a similar level of luciferase induction to that conferred by the wildtype element (5.4-and 5.9fold, respectively). These data suggest that the sequence of the LdNT3 loop, but not stem, is essential for purine-responsive repressor activity in L. donovani.
We noted two distinct blocks of conservation within the loop regions of LdNT3 stem-loop orthologs (labeled A and B in Figure 5A). We generated LdNT3 stem-loop variants in which blocks A and B were mutated independently and tested their activity at the LdNT4 locus. Disruption of either block resulted in a complete loss of regulation ( Figure 5B), indicating that these conserved regions are important for the repressor function of the LdNT3 stem-loop.
The LdNT3 loop differs from that of TbNT8.1 at just three positions ( Figure 6A). To determine if the inactivity of the TbNT8.1 stem-loop in L. donovani could be attributed to any one of these divergent bases, we generated a series of TbNT8.1 stem-loop mutants in which each variant position was changed to the corresponding base from the leishmanial ortholog. A TbNT8.1 stem with a wildtype LdNT3 loop (labeled as 1-3 in Figure 6B) served as a positive control for purine-responsive induction. Like the wildtype TbNT8.1 element, replacement of each of the three variant residues alone had no effect on luciferase activity, suggesting that species restriction is defined by multiple bases in the loop rather than any one individually. Similarly, simultaneous conversion of positions 1 and 2 to the corresponding LdNT3 bases failed to restore purine-responsive regulation (1,2 in Fig. 6B). Two paired-position mutants (1,3 and 2,3) conferred varying degrees of repression that translated to a respective ~2.1-fold and ~3.7-fold increase in luciferase activity under purine stress; however, neither fully recapitulated the robust ~7.7-fold induction observed in control cells where all three bases were changed to their LdNT3 counterparts. Thus, each of the three nonconserved bases is important for species specificity of the repressor stem-loops, and full repressor function in L. donovani depends on the sequence at all three positions. Interestingly, the sequence of the 2,3 loop variant ( Figure 6) corresponds to that from the orthologous T. cruzi transporter (Figure 2A), suggesting that the binding specificities of the cognate RNA binding proteins that associate with these stem-loops in L. donovani and T. cruzi has likely also diverged.
Regulation conferred by the LdNT3 stem-loop is likely mediated by a highly abundant transacting factor. The steady-state mRNA level of the Fluc-BSD transgene flanked by LdNT4 UTRs and harboring an LdNT3 stem-loop is approximately 75% lower than that of the same transgene expressed from the endogenous LdNT3 locus ( Figure 7B). Having demonstrated that the LdNT3 stem-loop is sufficient to confer purineresponsiveness to this lower-copy message, we next asked if the sequence could also mediate regulation of more abundant transcripts. The pRP vectors are a set of integrating rRNA promoter vectors generated in our laboratory that offer a range of incrementally different expression profiles in L. donovani. In all configurations, RNA polymerase I (Pol I) drives robust transgene transcription from the rRNA array. Graded expression is achieved using different combinations of UTRs that vary in their respective abilities to either promote or attenuate mRNA processing and stability (22). As depicted in Figure 7A, we introduced the LdNT3 stem-loop into pRP-LA and pRP-VH, low-and high-expressing vectors from the pRP suite. These constructs differ in the sequences of their 5'-UTRs but share an identical 3'-UTR derived from the L. major α-tubulin (LmTUB) intergenic region, facilitating the comparative study of the LdNT3 stem-loop at two steady-state transcript levels while eliminating the confounding variable of local genetic context. A Fluc reporter gene was inserted into the multiple cloning sites of these vectors and linearized constructs were integrated into a dualluciferase compatible cell line. As expected, the relative abundance of the Fluc transcript was substantially higher in these cells than when expressed from the endogenous LdNT3 locus, to over 90-and 300-fold in the case of stem-loopmodified pRP-LA and pRP-VH, respectively. Relative to control cell lines harboring an unmodified 3' UTR, the presence of a stem-loop in either vector substantially reduced Fluc mRNA levels, consistent with our previous observation that this sequence negatively affects mRNA stability ( Figure 7B). To test whether the LdNT3 stem-loop could mediate purine-responsive regulation of these higher-copy transcripts, cells were cultured for 48 hours in the presence or absence of purines and subjected to dual-luciferase analysis. Fluc activity was substantially reduced in both pRP-LA and pRP-VH control cells by 48 hours of purine stress, likely reflecting a general decrease in Pol I-mediated transcription of the rRNA array ( Figure 7C, left). This is supported by our previous observation that Pol I protein levels are significantly downregulated in purine-starved L. donovani (16). Interestingly, insertion of the LdNT3 stem-loop into the 3'-UTR of pRP-LA, but not pRP-VH, resulted in a moderate upregulation of luciferase activity upon starvation ( Figure 7C, right), which overcame the general starvation-induced reduction in expression from the pRP vectors. The fact that the LdNT3 stem-loop conferred purine responsive regulation to an mRNA ~90-fold more abundant than LdNT3 suggests that the trans-acting factors that associate with this element are in considerable excess of what is required solely for LdNT3 regulation. Given that the two vectors encode identical 3'-UTRs with the LdNT3 stem-loop inserted at the same position, the failure of the stem-loop to confer upregulation in the context of the pRP-VH construct is consistent with the possibility that the high level of mRNA expressed from this construct exceeded the availability of the cognate RNA binding protein.

DISCUSSION
Responding to the nutritional environment of the host is critical for successful parasitism. Our previous studies demonstrated that purine starvation invokes a robust nutrient stress response in Leishmania donovani that is characterized by a marked remodeling of the cellular proteome. We established that purine nucleoside and nucleobase transporters, including LdNT3, were highly upregulated by purine stress, and that this regulation was mediated, at least in part, via regulatory elements encoded in 5'-and/or 3'-UTRs of their mRNAs (14,16,18). In the current work, we have built on these observations by performing a detailed examination of the post-transcriptional and post-translational regulation of the LdNT3 purine nucleobase transporter in response to purine starvation.
Based on homology to a known purineresponse element from T. brucei (13), we identified a 33 nt stem-loop in the 3'-UTR of LdNT3 that serves to repress expression when purines are abundant. This is the first report of a defined nutrient stress-response element in Leishmania. Deletion of this element almost entirely ablated purine-responsive control by the LdNT3 UTRs. Whereas previous work established that both LdNT3 mRNA stability and translation are increased by purine starvation (16,18), we found that post-translational stability of LdNT3 is not affected by extracellular purine level. Thus, in the absence of a post-translational contribution, this singular element appears to be the primary factor mediating purine-responsive changes in LdNT3 abundance.
The putative stem-loop described in this work is conserved across a variety of dixenous and 2 M. H. Licon and P. Yates, unpublished observation. monoxenous kinetoplastids including multiple Leishmania species, T. brucei subspecies, and Trypanosoma cruzi (highlighted in Figure 2A) as well as Crithidia fasciculata, Leptomonas pyrrhocoris, and Blechomonas ayalai ( Figure S1). Conservation is particularly striking given that the relative position of the stem-loop varies substantially with respect to the stop codon and the 3'-UTRs of the orthologous transporters are otherwise poorly conserved. Thus, despite substantial expansion or contraction of the UTRs, this sequence has been maintained throughout evolutionary history, suggesting a strong selective pressure to maintain purine-responsive regulation. Interestingly, the stem-loop is absent from the orthologous purine transporter gene of the freeliving kinetoplastid Bodo saltans, which also lacks the capacity for de novo purine synthesis (23). This observation may indicate that a robust purine stress response is uniquely important to the parasitic lifestyle.
As was shown for the TbNT8.1 stem-loop in T. brucei (13), the LdNT3 stem-loop is sufficient to confer regulation to an otherwise purineunresponsive reporter in L. donovani. However, sequence context appears to be important, since only one of two positions into which the LdNT3 stem-loop was inserted supported repressor activity (Fig. 3). This likely reflects differences in the propensities of the sequences surrounding the insertion points to negatively impact folding and/or accessibility of the stem-loop. The sequences of the TbNT8.1 and LdNT3 stem-loops differ by only 20%. We were therefore surprised to find that the two elements are not functionally equivalent. Species-restriction was attributed to just three variant bases encoded within the loop. One plausible explanation for this restriction may be that that orthologous RBPs that bind to these repressor elements have different specificities. However, as the relevant trans-acting factor in T. brucei was not identified and we have yet to isolate a candidate RBP for the LdNT3 stem-loop, this hypothesis has yet to be directly tested.
Numerous genes are differentially expressed in response to purine stress, yet we were unable to identify the LdNT3 purine-response element elsewhere in the L. donovani genome via bioinformatic analysis. 2 However, this does not preclude the possibility that LdNT3 is part of a purine-responsive regulon consisting of multiple genes under the control of a common RBP, as the ability of individual RBPs to interact with several disparate binding sites is well-documented (24, 25). Indeed, our observation that the stem-loop is sufficient to confer purine responsiveness to a transcript that is over 90-fold more abundant than LdNT3 strongly suggests that its binding partner is present in substantial excess of what is required for LdNT3 regulation, and may therefore play a role in regulating other genes within or outside of the purine stress response pathway. Despite this possibility, we suspect that the L. donovani purine stress response is likely mediated by multiple independent but intersecting pathways. For instance, each of the membrane purine transporters appears to be regulated by a unique combination of post-transcriptional regulatory mechanisms; the L. donovani purine nucleoside transporter 2 (LdNT2) is regulated solely at the translational level, while both mRNA abundance and translation are altered for purine nucleoside transporter 1 (LdNT1.1) and LdNT3 during purine stress (16). Moreover, we recently obtained evidence that the 3'-UTRs of LdNT1.1 and LdNT2 encode activator elements which, unlike the repressive stem-loop described here, promote expression when extracellular purines are depleted. 3 Future efforts to identify the proteins that associate with these cis-acting elements will help to unravel the complexities of the purine stress response pathway and may uncover novel targets for therapeutic intervention.

EXPERIMENTAL PROCEDURES L. donovani culture.
All cell lines described herein were generated from the L. donovani 1S-2D clonal subline LdBob, originally obtained from Dr. Stephen Bevereley (26). LdBob promastigotes were routinely maintained at 26 °C in 5% CO2 and cultured in Dulbecco's Modified Eagle-Leishmania (DME-L) medium supplemented with 5% SerumPlus TM (SAFC BioSciences/Sigma Aldrich, St. Louis, MO; a purine-free alternative to standard FBS), 1mM Lglutamine, 1x RPMI vitamin mix, 10uM folic acid, 50 ug/ml hemin, and 100 uM hypoxanthine as a purine source. For general culture maintenance, blasticidin and puromycin were used at 30 ug/ml and 25 ug/ml, respectively. To elicit purine starvation, logarithmically growing cells were pelleted via centrifugation (5000 x g for 5 min), washed once in DME-L lacking hypoxanthine but containing all other media supplements, and resuspended at a density of 2 x 10 6 cells/ml in either purine-replete or purine-free medium.

Luciferase constructs and cloning.
All gene targeting constructs were generated using the multi-fragment ligation approach described in (27) and depicted in Figure S2. To genetically fuse Fluc the LdNT3 CDS (constructs S2B and S2C), a BSD-2A-Fluc transgene was provided by donor vector pCRm-coBSD-2A-Fluc. 4 The Fluc-BSD reporter used for gene replacement constructs ( Figure S2D) was donated by pCRm-luc2-BSD (Genbank Accession number KF035118.1). 5'-and 3'-targeting sequences for integration into either the LdNT3 or LdNT4 locus were PCR amplified from genomic DNA with Phusion High Fidelity DNA polymerase (New England Biosciences, Ipswitch, MA) using primers listed in Table S1. For construct assembly, all vector components were digested with SfiI (or AlwNI, where indicated), gel-purified, and combined in a single ligation step as depicted in Figure S2A.
To generate LdNT3 stem-loop-deficient mutants, the LdNT3-targeting construct shown in Figure S2D was modified with the QuikChange Site-directed Mutagenesis Kit (Stratagene, Lajolla, CA) using the primers listed in Table S2. To insert the LdNT3 stem-loop and variations thereof into the LdNT4 3'-UTR, the LdNT4 version of construct S2D was subjected to whole-plasmid PCR amplification via Phusion polymerase using the primers listed in Table S3. PCR products were DpnI-treated to eliminate template plasmid and circularized with NEBuilder HiFi DNA Assembly Master Mix (New England Biosciences, Ipswitch, MA) according the manufacturer's instruction.
All primers used to modify pRP vectors are listed in Table S4. To integrate the firefly luciferase gene into the rRNA array, Fluc was amplified from pCRm-luc2-BSD and cloned into the SfiI sites of pRP-LA and pRP-VH (22). Insertion of the LdNT3 stem-loop into pRP-LA and pRP-VH was accomplished in a step-wise fashion. First, the vectors were subjected to whole-plasmid amplification to introduce two nonidentical BstXI sites into their shared 3'-UTRs (see primer sequences for detail). A version of the LdNT3 stemloop was synthesized with flanking BstXI and PCR primer binding sites (Genscript, Piscataway, NJ) and inserted into the BstXI sites of the modified pRP vector.

Transfections
To generate dual-luciferase cell lines, an LdBob derivative expressing Rluc from the endogenous UMPS locus was used as a recipient for all vector transfections (18). Transfections of midlog stage promastigotes were performed with ~3ug SwaI-linearized plasmid DNA using the highvoltage electroporation protocol described by Robinson and Beverley (28). Immediately following electroporation, cells were transferred into 5 mL of complete DME-L and 200 ul was added to the first column of wells on a 96 well plate and subjected to 2-fold serial dilution to derive independent clones. Transfections were incubated overnight at 26 °C in 5% CO2. Selection was initiated the following day by adding 100 ul of 2X blasticidin (60 ug/ml) or, for integration of pRP-LA and pRP-VH vectors harboring NEO, G418 (50 ug/ml) to each well. Proper integration of the constructs was verified via PCR for all clones.

Dual-Luciferase analysis.
Firefly and Renilla luciferase activities were assessed using the Dual-Glo Luciferase Assay System from Promega. Analyses were performed using 35 ul of cell culture in white polystyrene 96well half-area plates (Corning, Amsterdam) as described in the product technical manual. For each incubation step, plates were protected from light and shaken for 10 minutes at room temperature on an orbital shaker. Luminescence was measured using a Veritas Microplate Luminometer (Turner BioSystems, Sunnyvale, CA).

RTq-PCR Analyses.
Total cellular RNA was isolated from 5 x 10 6 log-stage parasites using the Qiagen RNeasy Plus Micro Kit following the protocol for animal and human cells. Cell lysates were disrupted using a QIAshredder spin column. To eliminate contaminating genomic DNA, RNA samples were subjected to DNaseI digestion using the TURBO DNA-free kit. First-strand cDNA synthesis was performed using 1ug of RNA template with the High Capacity cDNA Reverse Transcription Kit. cDNA samples were subsequently diluted so as to reduce the total input RNA to 6 ng per ul. Dyebased real-time qPCR was performed with NEB Luna Universal qPCR Master Mix according to the manufacturer's instructions using 2 ul (12 ng) of diluted cDNA. Previously validated PCR primers (16) are listed in Table S5. Reactions were run on an Applied Biosystems StepOnePlus instrument using the "Fast" ramp speed and the following thermocycling parameters: 95 °C for 60 seconds; 40 cycles of denaturing at 95 °C for 15 seconds followed by a 30 second extension at 60 °C. A final melt curve step was included to verify the specificity of amplification. The relative abundance of the Fluc transcript expressed from various genetic loci was determined using the comparative CT (ΔΔCT) method as described (16).   Figure 3A. Bars   A) Differential high-level Fluc expression achieved using integrating rRNA promoter vectors. 5'-UTRs are derived from the upstream IGRs of the T. brucei procyclic acidic repetitive protein and Crithidia fasciculata phosphoglycerate kinase B gene in pRP-LA and pRP-VH, respectively. Abbreviations and symbols: term = putative rRNA terminator sequence; NEO = neomycin resistance gene; black boxes = rRNA promoter; vertical hash marks = 64 base repeats. Black arrows indicate the direction of transcription by RNA polymerase I. Grey shading indicates homology for targeted integration. B) Relative mRNA abundance for Fluc expressed from either the endogenous LdNT4 locus or the rRNA locus via pRP-LA and pRP-VH vectors was determined by RTq-PCR. All data were normalized to the endogenous control UMPS. The mean and standard deviation from two biological replicates is shown for each analysis. Transcript level is represented relative to Fluc expression from the endogenous LdNT3 locus. C) Fluc activity from the pRP-LA and pRP-VH vectors after 48 hours of culture in the presence or absence of purines. End points represent the mean of assays performed in technical duplicate for three independent clones of each cell line.