Trypanosoma brucei PRMT1 Is a Nucleic Acid Binding Protein with a Role in Energy Metabolism and the Starvation Stress Response

Trypanosoma brucei infection causes human African trypanosomiasis, also known as sleeping sickness, a disease with a nearly 100% fatality rate when untreated. Current drugs are expensive, toxic, and highly impractical to administer, prompting the community to explore various unique aspects of T. brucei biology in search of better treatments. In this study, we identified the protein arginine methyltransferase (PRMT), TbPRMT1, as a factor that modulates numerous aspects of T. brucei biology. These include glycolysis and life cycle progression signaling, both of which are being intensely researched toward identification of potential drug targets. Our data will aid research in those fields. Furthermore, we demonstrate for the first time a direct association of a PRMT with nucleic acids, a finding we believe could translate to other organisms, including humans, thereby impacting research in fields as distant as human cancer biology and immune response modulation.

IMPORTANCE Trypanosoma brucei infection causes human African trypanosomiasis, also known as sleeping sickness, a disease with a nearly 100% fatality rate when untreated. Current drugs are expensive, toxic, and highly impractical to administer, prompting the community to explore various unique aspects of T. brucei biology in search of better treatments. In this study, we identified the protein arginine methyltransferase (PRMT), TbPRMT1, as a factor that modulates numerous aspects of T. brucei biology. These include glycolysis and life cycle progression signaling, both of which are being intensely researched toward identification of potential drug targets. Our data will aid research in those fields. Furthermore, we demonstrate for the first time a direct association of a PRMT with nucleic acids, a finding we believe could translate to other organisms, including humans, thereby impacting research in fields as distant as human cancer biology and immune response modulation. KEYWORDS PRMT, RNA binding proteins, Trypanosoma brucei, arginine methylation, metabolism regulation, stress response T rypanosoma brucei is a parasitic protozoan causing African sleeping sickness in sub-Saharan Africa. An estimated 70 million people are at risk of the infection, and WHO estimates roughly 20,000 new cases per year when likely underreporting is taken into account (1). Furthermore, animal trypanosomiasis in the African region constitutes a large economic burden. It is estimated that dealing with trypanosomiasis would result in a benefit of approximately 2.5 billion USD to livestock keepers in affected regions over a 20-year period (2). The parasite is transmitted between the mammalian hosts via an insect vector, the tsetse fly (Glossina spp.). Throughout its life cycle, T. brucei changes both its morphology and physiology to adjust to nutritional and immunological conditions encountered in the hosts. The bloodstream form (BF) that thrives in mammalian blood utilizes glycolysis, compartmentalized in a specialized organelle called a glycosome, as the main energy source (3). BF cells employ a quorum sensing mechanism to detect a high parasite load and transform to a nondividing stumpy stage that is preadapted to life in the insect vector (4). Once taken up by the fly, parasites progress through the life cycle, further changing their physiology. The procyclic form (PF) inhabiting the fly's midgut turns to proline degradation coupled to the TCA cycle to cope with the lack of glucose in its environment (5). These changes are reflected in the size of the parasite's single mitochondrion, which in PF takes up much of the cytoplasmic space, as well as in the utilization of oxidative phosphorylation, which is almost exclusively active in PF. The changes T. brucei undergoes through its life cycle are almost solely controlled at the posttranscriptional level, since T. brucei utilizes polycistronic transcription of functionally unrelated genes and subsequently generates individual mRNAs through the processes of 5= trans-splicing and 3= cleavage and polyadenylation (6). The major means of gene regulation are embodied in the mRNA binding proteome, which in turn is regulated by a multitude of mechanisms, including posttranslational modifications (PTM). Protein arginine methylation is a PTM that disproportionally targets RNA binding proteins in T. brucei as well as in mammals (7)(8)(9). Arginine methylation, which in T. brucei affects about 15% of the proteome, is catalyzed by protein arginine methyltransferases (PRMTs) that can be classified into three types based on the end products of their catalytic activities (7,8,10). All three types can catalyze the formation of -N G -monomethylarginine (MMA). While type III PRMTs are limited to this modification, type I PRMTs can catalyze formation of -N G ,N G -asymmetric dimethylarginine (ADMA), and type II PRMTs create -N G ,N= G -symmetric dimethylarginine (SDMA). T. brucei harbors four PRMTs that represent all three types and engage in a functional interplay (11).
T. brucei PRMT1 (TbPRMT1) is a type I PRMT that we previously showed catalyzes the majority of ADMA formation in vivo (12). TbPRMT1 influences the functions of proteins involved in RNA editing and controls the mRNA-stabilizing and -destabilizing functions of the RNA binding protein DRBD18 in PF T. brucei (12)(13)(14)(15)(16). However, more global impacts on cell function have not been investigated, and TbPRMT1 function has not been examined in BF T. brucei. TbPRMT1 structure is highly unusual since, unlike other characterized type I PRMTs that form homodimers or homomultimers, it functions as a heterotetramer of an enzymatic subunit (ENZ) and a catalytically inactive PRMT paralog termed prozyme (PRO) (13,(17)(18)(19)(20). The two TbPRMT1 subunits mutually stabilize each other on the protein level, and while the PRO subunit does not carry out catalysis, it is indispensable for TbPRMT1 function.
Here, we examine the in vivo role and in vitro properties of TbPRMT1 using a mouse model and biochemical, cell biological, and global proteomic strategies in BF and PF T. brucei. Our results show that although TbPRMT1 is not strictly needed for BF T. brucei growth in culture, the protein contributes to T. brucei virulence in an animal model. We further show that in the absence of TbPRMT1, the BF parasites downregulate enzymes involved in glycolysis and upregulate pathways that utilize alternative energy sources.
We quantified changes in the mRNA-bound proteome and identified several proteins whose association with mRNA is significantly altered in the TbPRMT1-depleted background. In the attempt to identify TbPRMT1 in vivo substrates, we noted an overrepresentation of stress-related proteins associating with TbPRMT1. We confirmed the biological significance of this finding by demonstrating a defect in mRNA granule formation during nutritional stress in PF cells depleted for TbPRMT1. Finally, we show that TbPRMT1 itself is able to associate with nucleic acids, which is a completely novel feature in this class of enzymes. Thus, the present studies reveal that TbPRMT1 plays significant roles in trypanosome virulence, metabolism, and RNA biology.

RESULTS
Knockout of the TbPRMT1 enzymatic subunit leads to growth retardation and decreases parasite virulence. We began by establishing whether TbPRMT1 is essential for T. brucei survival in the mammalian host. As an RNAi-based knockdown had no effect on BF growth in vitro (data not shown), we generated a TbPRMT1 knockout (KO) cell line in BF T. brucei to unequivocally determine whether the enzyme plays a role in virulence. Two alleles of the TbPRMT1 ENZ subunit were replaced with blasticidin and puromycin resistance cassettes, whose incorporation into the genome was confirmed by PCR on genomic DNA using both TbPRMT1 untranslated region (UTR)-based and open reading frame (ORF)-based primers (Fig. 1A). Growth of wild-type (WT) and KO cell lines was then measured over the course of 11 days, and a mild but reproducible growth phenotype was observed (Fig. 1B). To ascertain whether this phenotype could be attributed to the loss of TbPRMT1 or whether it should be ascribed to cell-linespecific differences, we complemented the KO with the ENZ subunit of TbPRMT1 exogenously expressed under doxycycline control. We expressed the exogenous ENZ for 2 weeks and then monitored growth over a 3-day period. We calculated doubling times of WT, ENZ KO, and ENZ KOϩAB (add-back) cell lines and determined that the mild growth phenotype could indeed be attributed to the loss of TbPRMT1 (Fig. 1C). Next we confirmed that the exogenous ENZ also restores the arginine methyl landscape of BF trypanosomes (Fig. 1D). We probed WT, ENZ KO, and ENZ KOϩAB cell lysates with anti-ENZ, anti-ADMA, and anti-MMA antibodies. We saw that loss of ENZ leads to a steep increase in proteins bearing the MMA mark, a phenomenon previously observed in both mammals and T. brucei (11,21). We also observed a modest decrease in proteins recognized by the anti-ADMA antibody. It is important to note that the anti-ADMA antibody recognizes only a small number of all proteins bearing the ADMA mark in the cell. In previous studies we found that while the observed decrease of proteins recognized by this antibody is modest, the increase of proteins bearing the MMA mark is quite striking (11). Upon exogenous expression of ENZ, the ADMA profile fully corresponded to the WT landscape. The MMA profile mostly corresponded to the WT landscape. Therefore, we conclude that the changes in methyl landscape in the ENZ KO cell line can be attributed to the loss of TbPRMT1. Having established the ENZ KO strain, we infected mice with either WT or ENZ KO T. brucei to determine the impact of TbPRMT1 in a living system. We found that ENZ KO leads to a significantly prolonged life expectancy of T. brucei-infected mice compared to WT (Fig. 1E, P Ͻ0.05). Therefore, TbPRMT1 contributes to T. brucei virulence through an unknown mechanism, which may be connected to the lower growth rate of ENZ KO cells in vitro.
TbPRMT1 knockout leads to changes in metabolic enzyme expression and changes mRNA association of a small subset of proteins. RNA binding proteins are a major protein group targeted by PRMTs across the evolutionary spectrum (7)(8)(9)22). Although arginine methylation of RNA binding proteins has mostly been shown to alter protein-protein interactions, examples of altered affinity for RNA upon modulation of arginine methylation state have been reported (9,(22)(23)(24). Since RNA binding proteins constitute the major means of regulating the T. brucei proteome, we asked whether the lack of TbPRMT1 alters the association of certain proteins with mRNA. To this end, we isolated the mRNA-bound proteome from WT and ENZ KO cell lines and subjected the eluates to quantitative mass spectrometry. We anticipated that any observed changes in the mRNA-bound proteome could be attributed to either altered mRNA association or altered protein abundance in the cell. In order to discriminate between these two possibilities, we first established steady-state abundances of proteins in WT and ENZ KO cell lines by quantitative mass spectrometry analysis. We identified 4,856 proteins out of the 9,598 currently annotated in the TriTryp database as protein coding. We saw that 385 proteins decreased in abundance (Ͼ1.5-fold, P Ͻ 0.05) and 167 proteins increased their abundance (Ͼ1.5-fold, P Ͻ 0.05) in the KO compared to WT (see Table S1 in the supplemental material). We confirmed these changes by Western blotting for several proteins that did not change in abundance; MRB800, which decreased in abundance; and UMSBP1, which increased in abundance ( Fig. 2A and Fig. S1). Further analysis of the data set revealed alterations in the levels of proteins involved in various aspects of T. brucei biology such as cell division, DNA repair and replication, RNA synthesis and processing, translation, the ubiquitin system, and vesicular transport. We also noted numerous remarkable alterations in the abundances of proteins involved in energy metabolism, some of which suggested possible pathway coregulation and life cyclespecific regulation. For example, ENZ KO leads to upregulation of several enzymes in the proline degradation pathway ( Fig. 2B and Table 1). Interestingly, the mRNAs encoding all of the enzymes needed to obtain energy from the oxidation of proline to succinate are bound by a common RBP, DRBD3 (25). We identified significant overlap between DRBD3-bound transcripts and proteins upregulated in ENZ KO (P ϭ 0.009), suggesting that this RBP may mediate a subset of the changes observed in Table 1. The proline degradation pathway is typically utilized in PF T. brucei cells that lack access to glucose (5), leading us to examine other metabolic pathways whose utilization is altered during the T. brucei life cycle. We observed an upregulation of certain subunits of respiratory chain complexes and farnesyl pyrophosphate synthase, which is involved in synthesis of CoQ ( Table 1). The most prominent increase in abundance was noted in the citric acid cycle, namely, in the level of citrate synthase, which increased 10-fold. We also noted a marked decrease in enzymes participating in the major pathway of energy production in BF T. brucei: glycolysis. Specifically, hexokinase, the enzyme responsible for glucose commitment to glycolysis, decreased in abundance to about 50% of WT levels. Similar decreases were noted in the levels of glucose-6-phosphate isomerase, an enzyme working directly downstream of hexokinase, and in alanine aminotransferase. Alanine aminotransferase is responsible for conversion of pyruvate to alanine, a pathway that is extensively utilized in BF trypanosomes (26). We also noticed a decrease in several enzymes with putative roles in inositol metabolism (Table 1, "other metabolismrelated genes"). Three out of four T. brucei fatty acyl CoA synthetases and glycerol-3phosphate acyltransferase decreased in abundance, indicating a decrease in the rate of de novo lipid synthesis that utilizes free fatty acids and glycerol. Possibly related to TbPRMT1 Influences Metabolism and Stress Response ® these changes could be the roughly 50% decrease in aquaglyceroporin 2 (AQP2) channel abundance. AQP2 is a glycerol transporter residing in the T. brucei cell membrane that has been linked to drug resistance in African trypanosomes (27).
Overall, the observed changes in metabolism resembled those typically observed during the transition from BF to PF. However, when we compared the entire data set to proteins previously shown to be up-/downregulated in PF or BF, we did not see a significant overlap with those that change upon ENZ KO (28). We did, however, observe an increase in PAD2, a protein involved in conveying the differentiation signal in the BF-to-PF life cycle transition (29). PAD2 is absent from the long slender BF stage, increases in abundance in the stumpy stage, and remains on the surface once the cells progress to the PF stage. We wanted to verify the increase in PAD2 in BF upon ENZ KO through Western blotting using anti-PAD2 antibodies (29). We were unable to quantitate the increase due to the absence of PAD2 signal in our WT BF cells. However, we do clearly observe a detectable amount of PAD2 in ENZ KO cells, validating the increase in this key differentiation molecule (Fig. 2C). Overall, we conclude that the loss of ENZ leads to a dysregulation of T. brucei metabolism. This effect may be achieved through regulatory mechanisms that play a role in life cycle progression or alternatively may be a compensatory effect as the cells cope with the stress of ENZ loss.
Having established steady-state levels of proteins in our WT and ENZ KO cell lines, we proceeded to quantify changes in the mRNA-bound proteome. WT and KO cells were UV cross-linked, and mRNA was affinity purified using oligo(dT)-coated magnetic beads (30). The mRNA with bound proteins was eluted off the beads, and proteins were quantified by mass spectrometry (Table S2). We identified 234 proteins, including 129 out of 155 proteins that were previously identified as mRNA binding in BF T. brucei by Lueong et al. (30). We observed significant changes in the abundance of 16 proteins (Ͼ1.5-fold, P Ͻ 0.05) in ENZ KO compared to WT parasites ( Table 2). Of these, 14 proteins were also identified in our whole-cell proteome analysis (Table S1). Only the levels of RBP35 (Tb927.9.12360) and Nopp44/46-2 (Tb927.8.750) significantly changed on the whole-cell level, although eight proteins showed a trend that suggested a possible change but did not pass the significance criteria (Table 2). To verify the mRNA-bound proteome data set, we repeated the mRNA purification experiment in biological triplicate and detected selected proteins by Western blotting (Fig. 3A). Based on available antibodies, we chose to assay levels of ZFP3 (Tb927.3.710, unchanged), NOPP44/46-1 (Tb927.8.760, decreased), and aldolase (Tb927.10.5620, increased). The aldolase Western blot did not reveal a significantly increased association with mRNA in the ENZ KO cell line, although the quantification showed a trend consistent with a slight increase. This result was not surprising considering that, unlike ZFP3 and NOPP44/46-1 proteins, which are both involved in nucleic acid metabolism, aldolase is an extremely abundant metabolic enzyme. Thus, the increased mRNA association of aldolase in our experiment may be due to contamination of TbPRMT1 KO samples. In contrast, Western blot analysis confirmed our mass spectrometry results for ZFP3 and NOPP44/46-1, demonstrating unchanged levels for the former and a decrease of the latter to less than 30% of WT levels. Interestingly, we previously showed that NOPP44/ 46-1 is heavily decorated by ADMA, consistent with an important role for TbPRMT1 in its activity (7).
Having verified the data set, we more closely analyzed the proteins exhibiting altered mRNA association in ENZ-depleted cells. We first asked whether these proteins have commonalities in their subcellular localization by examining data from the TrypTag project. Interestingly, out of the 8 proteins tagged to date, 7 localize at least partially to cytoplasmic punctae in PF T. brucei (Tb927.10.4430, Tb927.9.4080, . We were also intrigued by the change in mRNA association of eIF4G5, its interacting protein G5-IP, and two members of the NOPP44/46 family, given the reported translation-related functions of these proteins. eIF4G5 forms a complex with eIF4E6 and G5-IP, a protein with similarity to capping enzymes. This complex has affinity for capped mRNAs, but its depletion does not significantly decrease the overall translation rate (31). NOPP44/46 is a trypanosome-specific protein family that bears extensive, life cycle-regulated tyrosine phosphorylation and has been implicated in ribosomal 60S subunit maturation (32). Since NOPP44/46-1 is known to shuttle between nucleus and cytoplasm, we first wanted to determine whether the decrease in NOPP44/46-1 mRNA binding could be due to mislocalization (33). We immunolocalized the protein but did not observe any change between the WT and ENZ KO cell lines (Fig. 3B). Given the potential roles of the above-mentioned proteins in translation, we next asked whether ENZ KO leads to an altered translation rate of all or a subset of proteins. We pulsed the cells with [ 35 S]methionine and separated the nascent proteins by SDS-PAGE. We detected a reproducible decrease in translation of an ϳ50-kDa and an ϳ38-kDa protein but did not see any overall translation defect (Fig. 3C). It is important to note that this assay lacks the resolution to visualize low-abundance proteins; therefore, it is quite possible that the number of proteins whose translation rate is affected exceeds the two easily visualized proteins. We conclude that the loss of TbPRMT1 alters the mRNA binding capacity of a small number of cytoplasmic proteins, which in turn may lead to specific impacts on mRNA translation. TbPRMT1 is necessary for an efficient T. brucei starvation stress response. To begin to distinguish secondary effects of TbPRMT1 depletion from the primary affected pathways, we need to determine possible TbPRMT1 substrates. Since our antibodies are not specific enough to immunoprecipitate either ENZ or PRO subunit, we attempted to purify exogenously expressed PRO-Myc-His-TAP (PRO-MHT) from BF T. brucei to identify associated proteins that may be substrates. We were unable to achieve sufficient expression for proteomic analysis in BF and so switched to PF. We successfully expressed PRO-MHT and purified associated proteins (see Table S3 in the supplemental material). Complexes were eluted by TEV protease cleavage to minimize contaminants, and cells in which PRO-MHT expression was not induced were used as a negative control. We refined our data set by eliminating all proteins that were identified by Ͻ2.5ϫ peptides over the negative control, using the replicate with lower peptide count for that specific protein. Applying this criterion, we identified enriched biological process GO terms for our data set. We observed a significant enrichment for proteins involved in metabolic processes and various levels of gene expression, findings that supported our previous data indicating TbPRMT1 involvement in regulation of T. brucei metabolism (Fig. 4A). Unsurprisingly, when we looked at molecular function GO terms,  (34). P value calculated using R p-hyper function.

TbPRMT1 Influences Metabolism and Stress Response
® the overwhelming majority of identified proteins fell into the RNA binding protein category (Fig. 4A). A single protein was identified in common between the proteins that changed their mRNA association and proteins that copurify with TbPRMT1: a nucleolar protein with unknown function (Tb927.9.10400). Although not identified as methylated on arginine residues in our proteome-wide screen (7,8; also unpublished data), the Tb927.9.10400 sequence contains multiple RGG motif repeats, which are typically targeted by arginine methylation, making Tb927.9.10400 a promising target for future studies. One finding that caught our eye was the significant overlap of proteins that copurify with TbPRMT1 and proteins that are enriched in T. brucei starvation granules (34) (Fig. 4B) (P ϭ 5.3 ϫ 10 Ϫ8 ) ( Table 3). This includes SCD6 (Tb927.11.550), a stressrelated protein that was previously reported to copurify with TbPRMT1 (35). Association of TbPRMT1 with starvation stress-related granules was particularly striking since the PRO subunit of TbPRMT1 itself was enriched in a recent proteomic analysis of stress granules (34). To investigate a possible role of TbPRMT1 in stress granule formation, we utilized a PF ENZ RNAi cell line that efficiently ablates PRO and ENZ protein levels when induced by doxycycline (11). Either uninduced or RNAi-induced cells were starved by a two-hour incubation in PBS, and mRNA was subsequently visualized by oligo(dT) fluorescent in situ hybridization (FISH) (Fig. 5A). We classified the cells as either stressed (clearly containing RNA granules), intermediate (some granulation might be present), or normal (mRNA distributed evenly throughout cytoplasm) (Fig. 5B, lower panel). We observed a statistically significant decrease in cells that were able to form stress granules upon TbPRMT1 depletion (Fig. 5B, upper panel). Together, these data demonstrate a role for TbPRMT1 in starvation stress granule formation in PF T. brucei and suggest potential substrates that may contribute to this effect.
TbPRMT1 directly binds to nucleic acid. The effect of TbPRMT1 on stress granule formation led us to investigate the possibility that TbPRMT1 might itself be able to associate with RNA. This notion was supported by a recently reported mRNA-bound proteome study in which more PRO peptides were purified with the mRNA than in the negative control, although the significance criterion was not met (30). To investigate whether TbPRMT1 has RNA binding properties, we began by analyzing the association of TbPRMT1 with mRNA in vivo. We attempted to use our antipeptide TbPRMT1 antibodies; however, the results were inconclusive due to the interference of nonspecific bands (data not shown). Therefore, we utilized our PRO-MHT cell line to purify the mRNA-bound proteins. Due to the presence of the protein A moiety in the TAP tag, the secondary anti-rabbit IgG antibody strongly recognized the tagged protein (Fig. 6). We saw a clear association of PRO-MHT with mRNA as well as the presence of our positive control, DRBD18. Our negative control, p22, was not present in the final eluate. Therefore, we conclude that PRO associates with mRNA in vivo. To confirm that PRO TbPRMT1 Influences Metabolism and Stress Response ® directly associates with RNA, we turned to in vitro methods. Since our attempts to identify in vivo TbPRMT1 mRNA targets were unsuccessful, we utilized a 102-nt 32 Plabeled pBluescript SKϪ plasmid RNA. We generated recombinant TbPRMT1 heterotetramer by ENZ and His-PRO coexpression in Escherichia coli followed by nickel affinity purification of the complex as described previously (12). Individual TbPRMT1 subunits were expressed as His-ENZ or His-PRO separately and purified. We incubated the RNA with TbPRMT1, individual TbPRMT1 subunits, and other T. brucei PRMTs; UV cross-linked the samples; separated them by SDS-PAGE; and visualized RNA-bound complexes (Fig. 7A). We observed no RNA binding when only the ENZ TbPRMT1 subunit was present, although this result should be viewed with caution, since the ENZ subunit is difficult to purify by itself and is prone to aggregation. Importantly, we saw clear RNA binding of the PRO subunit both incubated alone with RNA and within the context of the entire TbPRMT1 heterotetramer. We also observed binding of TbPRMT7, a type III PRMT that was enriched in the published T. brucei mRNA-bound proteome to a similar degree as TbPRMT1 (30). TbPRMT6 was not found cross-linked to RNA, suggesting that in vitro RNA binding is not a default feature of all PRMTs. Next, we wanted to utilize another, more native method to confirm that TbPRMT1 binds RNA in vitro. We incubated increasing concentrations of TbPRMT1 or its PRO subunit with labeled RNA and resolved the reaction mixtures by native PAGE, resulting in an electrophoretic mobility shift (EMSA) of protein-bound RNA. We saw a distinct shift of RNA migration at the highest protein concentrations (2 and 4 M) (Fig. 7B). This suggests a relatively low affinity of TbPRMT1 for pBluescript RNA but confirms TbPRMT1 RNA binding ability. Some proteins exhibit affinity for specific polynucleotide sequences, and the in vivo mRNA binding activity of TbPRMT1 suggested it could exhibit high affinity for poly(A). To investigate whether TbPRMT1 has a preference for specific polynucleotides, we performed the in vitro RNA cross-linking assay using a 47-nt transcript of pBluescript sequence in the presence of increasing amounts of unlabeled poly(A), poly(C), poly(U), and poly(G) to determine their abilities to compete with the labeled probe for binding to TbPRMT1 (Fig. 7C). In these assays, PRO exhibited a preference for poly(U) and poly(G) but little binding to poly(A) or poly(C). Considering the low affinity of TbPRMT1 for RNA, we wanted to determine whether TbPRMT1 binds RNA specifically or whether the binding ability extends to all nucleic acids. We incubated TbPRMT1 with labeled RNA in the presence of unlabeled RNA or DNA competitor. We saw that the two competitors performed equally; therefore, we conclude that TbPRMT1 has nucleic acid binding ability (Fig. 7D). In another study (H. Hashimoto, L. Kafková, K. Jordan, L. K. Read, and E. W. Debler, submitted for publication), we determined that TbPRMT1 tetramerization is necessary for TbPRMT1 methylation activity. We thus wanted to test whether the tetramerization also influences TbPRMT1 RNA binding properties. We performed the RNA cross-linking assay on a battery of mutants that were unable to tetramerize (Fig. 7E). We observed that while tetramerization is not required for RNA binding, specific mutations in either the ENZ or PRO subunits of TbPRMT1 decrease the RNA binding potential of the protein complex. The finding that specific mutations are able to decrease TbPRMT1 RNA affinity strongly suggests that TbPRMT1 RNA binding is a specific and biologically relevant property of the enzyme, and it opens a window to further studies that will determine the residues directly involved in TbPRMT1 RNA binding.

DISCUSSION
Arginine methylation is a ubiquitous posttranslational modification that influences a broad spectrum of eukaryotic cellular processes (36)(37)(38)(39)(40)(41). The enzymes catalyzing this modification, PRMTs, are astonishingly well conserved throughout the eukaryotic kingdom. The enzymatic subunit of TbPRMT1, for example, shares 53% amino acid sequence identity with human PRMT1, a homology quite unusual between organisms with such evolutionary distance between them. In this study, we provide biochemical  Table at bottom of panel E describes the mutations. P or E in the superscript designates whether the mutation was in the PRO or ENZ subunit, respectively.
TbPRMT1 Influences Metabolism and Stress Response ® and cell biological characterization of TbPRMT1. We show that in BF T. brucei, TbPRMT1 promotes virulence in the animal model, influences energy metabolism, and changes mRNA association of a small subset of proteins. In insect-stage parasites, loss of TbPRMT1 significantly hinders the parasite's ability to form mRNA-containing cytoplasmic granules as a response to starvation stress. Furthermore, we show that TbPRMT1 is itself able to bind nucleic acids both in vitro and in vivo.
The virulence decrease of ENZ KO trypanosomes is interesting in the context of a phenotype observed in Leishmania major, a parasite closely related to Trypanosoma species. In contrast to what we observed in TbPRMT1 KO parasites, deletion of the LmjPRMT7 gene led to an increase in virulence, while overexpression caused decreased lesion progression in a mouse model (42). PRMTs are well known to engage in interplay in both T. brucei and human cells (11,21). Depletion of PRMT1 in both cell types causes a massive increase in MMA, which in T. brucei PF cells is attributed to TbPRMT7 activity (11). While the opposing effects of downregulating the two proteins could be a coincidence, especially considering that the data originated from different species, it is plausible that the virulence decrease we observe in the absence of TbPRMT1 results from increased TbPRMT7 activity.
Trypanosomes extensively modulate their energy metabolism throughout their life cycle as well as in response to available nutrients (43,44). In ENZ KO cells, we observed a number of alterations in metabolic enzyme abundance that are reminiscent of metabolic changes observed during T. brucei life cycle progression, namely, a decrease in glycolytic enzymes and an increase in components of the citric acid cycle and the proline degradation pathway. This prompted us to assess the ability of ENZ KO cells to undergo citrate-/cis-aconitate-induced in vitro differentiation from BF to PF. We did not observe any difference in the rate at which the cells acquired a procyclin coat or lost VSG coating compared to WT parasites (data not shown). This result could suggest that the observed changes are a result of slow growth rather than a reflection of interference with life cycle progression regulation (45,46). However, it is important to note that the citrate/cis-aconitate differentiation method forces the parasites to progress from BF to PF, skipping the stumpy stage, which our laboratory strain is not able to produce. Both the observed ϳ50% decrease in mRNA binding of HYP2 (Tb927.9.4080), a protein identified in a screen aimed at factors driving stumpy formation, and the increase in PAD2 protein levels in ENZ KO cells suggest possible TbPRMT1 involvement in BF to stumpy-stage progression (4). Furthermore, the observed defect in starvation granule formation in TbPRMT1-depleted cells hints at a possible defect in a life cycle progression through impaired response to the nutrient-poor environment that is naturally found in fly midgut and salivary glands. It is known that the differentiation of Trypanosoma cruzi, a close relative of T. brucei, from proliferative epimastigotes to infective growth-arrested metacyclic trypomastigotes requires nutritional stress as one of the signals (47,48). To better understand the role of TbPRMT1 in the life cycle progression of T. brucei, it will be necessary to generate a pleomorphic T. brucei cell line lacking TbPRMT1 and observe the ability of these cells to progress through the natural life cycle in the animal and fly hosts.
To begin understanding the direct effects of TbPRMT1-mediated methylation, it is important to identify TbPRMT1 substrates. Some of the observed effects are likely caused by regulation of RNA binding proteins either by direct methylation or by methylation of their binding partners. Our data show changes in the mRNA association of several RNA binding proteins. PUF1 and Tb927.9.10400 exhibit increased association with RNA in the absence of TbPRMT1. PUF1 depletion causes only minor changes in PF and BF gene expression, although this was investigated only by the relatively insensitive microarray approach, and the role of PUF1 in other T. brucei life cycle stages was not investigated (49). PUF1 is unlikely to be a direct target of TbPRMT1, since our proteome-wide screens did not identify arginine methylation of PUF1, nor was PUF1 found associated with TbPRMT1 (7, 8; unpublished data). Tb927.9.10400 is a nucleolar protein of unknown function (50). Despite the absence of Tb927. 9.10400 in our proteome-wide methylation screen, the protein sequence contains numerous RGG repeats that are typically targeted by arginine methylation. Furthermore, Tb927.9.10400 was in the top 5% of PRO-MHT-interacting proteins based on the spectral count, making it a promising candidate for future studies.
None of the proteins that decreased their mRNA association upon ENZ KO were found interacting with TbPRMT1. Nevertheless, the decrease in HYP2 mRNA binding that we observed in ENZ KO cells is intriguing. HYP2 contains a DksA zinc finger, a motif involved in nutritional status and quorum sensing responses in prokaryotes, and has been identified as a part of a quorum sensing signaling cascade that plays a role in stumpy-stage formation in T. brucei (4). The HYP2 protein sequence contains a number of arginines in the RG context, making it a potential target of arginine methylation. It is also possible that the HYP2-mRNA interaction is indirectly affected by TbPRMT1 activity. HYP11 (Tb927.11.2250) was also implicated to play a role in stumpy formation (4,51). HYP11 bears an ADMA mark, copurifies with TbPRMT1, has RNA binding properties, and is enriched in stress granules (4,30,34), strongly suggesting that HYP11 is a primary TbPRMT1 target. While HYP11 mRNA binding is not changed in our screen, it is possible that arginine methylation of HYP11 influences HYP2 mRNA binding or changes the composition of an HYP11-containing ribonucleoprotein, ultimately altering mRNA fate. It is important to note that our mRNA binding screen is likely biased toward identification of mRNA association changes in proteins with a narrow specificity toward a small subset of mRNAs, since changes in specificity of proteins with a large pool of bound mRNAs would not result in a statistically significant change in overall mRNA binding.
Other regulators of mRNA fate that could be affected by arginine methylation include DRBD3 and SCD6. DRBD3 is essential in PF cells and acts as a stabilizing factor of a small subset of mRNAs, some of which undergo stage-specific regulation (52). Considering that levels of DRBD3 remain constant between PF and BF cells, it is possible that stage-specific regulation is achieved through PTM decoration (52). Interestingly, among DRBD3 targets are the enzymes constituting the proline degradation pathway, several of which are upregulated in ENZ KO cells. DRBD3 was annotated as a posttranscriptional activator based on a high-throughput tethering assay (53). DRBD3 was not found methylated in our screen; however, it does harbor an RGG motif that is typically targeted by PRMTs. Starvation stress alters the composition of DRBD3-associated ribonucleoprotein complexes and causes DRBD3 to localize into cytoplasmic granules (54). Among DRBD3-interacting proteins are PABP1, PABP2, and NOT1, all of which copurify with TbPRMT1 and are also overrepresented in starvation stress granules (54). It is plausible that DRBD3 function is regulated through methylation of these proteins. Furthermore, DRBD3 appears as a possible mechanistic link between stage-specific metabolism regulation and starvation stress response. SCD6/RAP55 is a translational repressor with a function in stress granule formation in yeast and mammals (55)(56)(57). SCD6 function is modulated by HMT1/PRMT1 in yeast and mammals, and in T. brucei SCD6 strongly interacts with TbPRMT1 and is methylated on three arginine residues (35,(58)(59)(60)(61). Unlike SCD6 in other organisms, TbSCD6 does not seem to be necessary for formation of mRNA-containing granules under starvation conditions (35). However, TbSCD6 is a component of processing bodies, which are constitutive components of cytoplasmic mRNA metabolism in trypanosomes with a putative role in translational repression (62). The tight association of TbSCD6 with TbPRMT1 suggests that TbSCD6 is likely to play a role in some of the changes we observe on the proteome level.
TbPRMT1 could also be involved in directly regulating the functions of metabolic enzymes in the affected pathways. Glycolysis is the main pathway that provides energy for BF T. brucei cells (63). Upon knockout of ENZ, we see approximately a 50% decrease in hexokinase 2 (TbHK2) and phosphoglucose isomerase levels. We also noted a 40% decrease in levels of alanine aminotransferase, an enzyme responsible for conversion of pyruvate to alanine, a pathway that is extensively utilized in BF trypanosomes (26). Together, these data suggest a decrease in glycolytic flux, and direct measurement of the impact of TbPRMT1 knockout on the glycolytic rate is a future focus. Hexokinase catalyzes a virtually irreversible reaction that commits glucose to glycolysis. T. brucei TbPRMT1 Influences Metabolism and Stress Response ® carries two genes with hexokinase homology that are 98% identical at the amino acid level, TbHK1 and TbHK2. Unlike in other eukaryotes where hexokinases function as dimers, TbHKs form hexameric complexes, and they were isolated from cells as a heterohexamer containing an unknown ratio of TbHK1 and TbHK2 (64). T. brucei further differs from mammals in that TbHK is not inhibited by its product, glucose-6-phosphate (65). While TbHK2 is not active on its own, it gains activity when complexed with inactive TbHK1 (66). This complex then becomes amenable to inhibition by pyrophosphate, a molecule that has no effect on the TbHK1 homohexamer (66). These findings suggest that the two enzymes have distinct regulatory mechanisms that differ from the mammalian hexokinase regulation system. A peptide that could originate from either TbHK1 or TbHK2 (SKYRFVLPTTKFDLD) was shown to bear arginine dimethylation in our proteome-wide screen; therefore, TbPRMT1 is likely to contribute to this intricate and unusual pathway for regulation of glycolysis.
We show here that TbPRMT1 has nucleic acid binding properties and associates with mRNA in vivo. Future studies will be focused on determining the functional relevance of this property. One possibility is that TbPRMT1 can methylate RNA as well as arginines. TbPRMT1 shares features with the human RNA methyltransferase Mettl3/Mettl14. Mettl3/14 form a heterodimer in which Mettl3 is a catalytically active subunit while Mettl14 plays a critical structural role necessary for substrate recognition (67). Once in vivo RNA targets of TbPRMT1 are identified, they can be tested as potential substrates. Another possibility is that RNA binding plays a role in defining the appropriate topology of TbPRMT1 in respect to RNA-/DNA-bound substrates, as in the case of human lysine SET1 methyltransferase (68). Indeed, nucleic acid binding may be a conserved property of PRMT1 enzymes rather than a feature specific to T. brucei. PRMT1 was identified among the mRNA-associated proteins in two independent mRNA-bound proteome studies using human hepatocytic HuH-7 cells and human embryonic kidney HEK293 cells (69,70). Interestingly, human PRMT1 localizes to cytoplasmic granules upon arsenite treatment, suggesting further similarities between the two enzymes (71). Neither TbPRMT1 nor HsPRMT1 contains a known RNA binding domain. We identified specific residues in TbPRMT1 that, when mutated, attenuate its RNA binding activity, suggesting it may be possible to create a nucleic-acid-binding-deficient mutant that could in turn be used to further probe the biological significance of this feature.
In summary, we demonstrate here that TbPRMT1 affects T. brucei virulence, metabolism, and ability to cope with starvation stress. We identified a subset of proteins that change their mRNA association upon TbPRMT1 depletion, providing new directions for examining the role played by arginine methylation in regulation of RNA binding proteins. Furthermore, we established that TbPRMT1 itself has the ability to associate with nucleic acids, a finding that may have more widespread implications among PRMT1 homologs. We identified candidate TbPRMT1 targets that play a role in a broad spectrum of processes, extending the impact of this work to numerous fields of study, such as life cycle progression, translation, stress response, and metabolism regulation. Future studies will be aimed at distinguishing primary effects of TbPRMT1 loss from downstream proteome changes and deciphering the role that nucleic acid binding plays in TbPRMT1's mechanism of action.

MATERIALS AND METHODS
For details including buffer compositions and antibody sources, please refer to Text S1 in the supplemental material.
T. brucei cell culture and cell lines. PF T. brucei strain 29-13 and all cell lines derived from this strain were grown in SM medium supplemented with 10% FBS. The BF SM427 strain was cultured in HMI-11 medium supplemented with 10% FBS. The ENZ KO cell line was created by replacing both ENZ alleles with drug resistance markers. The PRO-MHT cell line contains a tetracycline-inducible C-terminally Myc-His-TAP-tagged PRO construct derived from pLEW100. The PRO-mNG-TY cell line was created using pPOTv6-mNG according to the published protocol (72). The ENZ RNAi cell line was created as described in reference 11.
Statistical analysis. Statistical significance of doubling time differences was calculated using multiple-comparison one-way ANOVA (Fig. 1). Western blot differences were tested by Student's t test ( Fig. 2 and 3). GO term enrichment P values were calculated by Fisher's exact test with Bonferroni correction (Fig. 4A). P value of overlap of two protein data sets was determined by R p-hyper function (Fig. 4B). Significance of differences between amounts of granule-containing cells in uninduced and induced ENZ RNAi cell lines was determined by the chi-square test (Fig. 5). For proteomics data sets, statistical significance between groups was evaluated using a Student t test (Table S1 and S2).
Mouse infection assay. Fourteen female CD-1 mice (10 to 12 weeks; Charles River) were infected by intraperitoneal injection with 1 ϫ 10 5 bloodstream-form T. brucei parasites; half were TbPRMT1 ENZ knockout (ENZ KO; n ϭ 7) or SM parental strain (WT; n ϭ 7). The course of infection was monitored, and time to death was recorded. Experiments were carried out in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Clemson University.
Whole-cell proteome and mRNA-bound proteome. Approximately 4 ϫ 10 6 cells were harvested, and whole-cell proteomes were quantified as described in Text S1. mRNA-bound proteome purification was performed as described previously, with only minor modifications (30). The protein description in Table 2 includes posttranscriptional activator/repressor annotation that was established by a genomewide tethering screen (53).
Immunolocalization and fluorescence in situ hybridization. NOPP44/46-1 was immunolocalized using antibody against the protein (1:2,000 dilution). FISH was performed using an oligo(dT) probe fused to Alexa Fluor 488 in an overnight (O/N) hybridization protocol.
In vitro RNA binding, competition, and EMSA. A 102-nt RNA fragment of pBluescript SKϪ plasmid sequence was in vitro transcribed. A UV-cross-linking-based RNA binding assay was performed as described previously with minor modifications (73). PRMT molarity was determined per dimer in all cases. Affinity for different polynucleotides was determined by adding a mass excess of polynucleotide into the reaction mixture prior to the addition of the protein. To perform EMSA, a 47-nt RNA fragment of the pBluescript SKϪ plasmid was transcribed and incubated with a recombinant protein. Reaction mixtures were resolved on native PAGE.