key: cord-0004606-h2xdkoyq
authors: Kieft, Rudo; Zhang, Yang; Marand, Alexandre P.; Moran, Jose Dagoberto; Bridger, Robert; Wells, Lance; Schmitz, Robert J.; Sabatini, Robert
title: Identification of a novel base J binding protein complex involved in RNA polymerase II transcription termination in trypanosomes
date: 2020-02-21
journal: PLoS Genet
DOI: 10.1371/journal.pgen.1008390
sha: a66e89a598f7195873f5556ec8829a53cbf6f32b
doc_id: 4606
cord_uid: h2xdkoyq

Base J, β-D-glucosyl-hydroxymethyluracil, is a modification of thymine DNA base involved in RNA Polymerase (Pol) II transcription termination in kinetoplastid protozoa. Little is understood regarding how specific thymine residues are targeted for J-modification or the mechanism of J regulated transcription termination. To identify proteins involved in J-synthesis, we expressed a tagged version of the J-glucosyltransferase (JGT) in Leishmania tarentolae, and identified four co-purified proteins by mass spectrometry: protein phosphatase (PP1), a homolog of Wdr82, a potential PP1 regulatory protein (PNUTS) and a protein containing a J-DNA binding domain (named JBP3). Gel shift studies indicate JBP3 is a J-DNA binding protein. Reciprocal tagging, co-IP and sucrose gradient analyses indicate PP1, JGT, JBP3, Wdr82 and PNUTS form a multimeric complex in kinetoplastids, similar to the mammalian PTW/PP1 complex involved in transcription termination via PP1 mediated dephosphorylation of Pol II. Using RNAi and analysis of Pol II termination by RNA-seq and RT-PCR, we demonstrate that ablation of PNUTS, JBP3 and Wdr82 lead to defects in Pol II termination at the 3’-end of polycistronic gene arrays in Trypanosoma brucei. Mutants also contain increased antisense RNA levels upstream of transcription start sites, suggesting an additional role of the complex in regulating termination of bi-directional transcription. In addition, PNUTS loss causes derepression of silent Variant Surface Glycoprotein genes involved in host immune evasion. Our results suggest a novel mechanistic link between base J and Pol II polycistronic transcription termination in kinetoplastids.

a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Termination of RNA polymerase II (Pol II) transcription of mRNAs is a tightly regulated process where the polymerase stops RNA chain elongation and dissociates from the end of the gene or transcription unit. However, the underlying termination mechanism is not fully understood. Pol II termination of most protein-encoding genes in eukaryotes is tightly linked to the processing of the nascent transcript 3' end (reviewed in [1] ). This association ensures complete formation of stable polyadenylated mRNA products and prevents the elongating Pol II complex from interfering with transcription of downstream genes. Transcription through the polyadenylation site results in an exchange of transcription factors, resulting in the regulation of the elongation-to-termination transition, in an ordered series of events: 1) dissociation of the elongation factor Spt5, 2) Pol II pausing, 3) changes in phosphorylation status of Pol II C-terminal domain (CTD), which promotes 4) recruitment of cleavage factors and termination factors, 5) transcript cleavage and 6) termination by the 5'-3' 'torpedo' exoribonuclease XRN2/Rat1. Critical to this process is the regulation of protein phosphorylation by the major eukaryotic protein serine/threonine phosphatase, PP1. As recently demonstrated in the fission yeast Schizosaccharomyces pombe (S. pombe), the PP1 phosphatase Dis2 regulates termination by dephosphorylating both the Pol II-CTD as well as Spt5 [2] . This PP1-dependent dephosphorylation allows the efficient recruitment of the termination factor Seb1 as well as decreased Spt5 stimulation of Pol II elongation that enhances the ability of the torpedo exoribonuclease to catch up and destabilize the elongation complex. Transcription termination was also affected in the S. cerevisiae PP1 homologue Glc7 mutant [3] . PP1 action is modulated through the formation of heteromeric complexes with specific regulatory subunits [4] . These regulatory protein subunits regulate PP1 by targeting the protein to specific subcellular compartments, to particular substrates, or reduce its activity towards potential substrates. The vast majority of regulators bind PP1 via a primary PP1-binding motif, the RVxF motif [4] [5] [6] . One of the key PP1 regulatory proteins in the nucleus is the PP1 nuclear targeting subunit (PNUTS) [7, 8] . PNUTS is a multidomain protein that contains the canonical PP1 RVxF interaction motif, an N-terminal domain that interacts with the DNA binding protein Tox4 and a domain near the C-terminus that interacts with Wdr82 [9] . This stable multimeric complex in humans is named the PTW/PP1 complex. Targeting of the complex to chromatin is presumably due in part through associations with Tox4. While the function of Wdr82 in this complex is not known, it may mediate interactions with Pol II by recognizing Ser5-phosphorylated CTD, as it does when it is associated with the Set1 complex [10] . In yeast, PP1 is associated with PNUTS and Wdr82 homologs in APT, a subcomplex of the cleavage and polyadenylation factor [11] [12] [13] [14] , and deletion of PNUTS and Wdr82 caused termination defects at Pol II-dependent genes [15, 16] . In mammals, PNUTS and Wdr82 mutant cells have defects in transcription termination at the 3' end of genes and 5' antisense transcription at bi-directional promoters [17] .

Members of the Kinetoplastida order include the parasite Trypanosoma brucei that causes human and animal African trypanosomiasis. Kinetoplastids are protozoa with unique genome arrangements where genes are organized into polycistronic transcription units (PTU) that are transcribed by Pol II. Pre-messenger RNAs (mRNA) are processed to mature mRNA by coupled 5' RNA trans-splicing and 3' polyadenylation [18] [19] [20] [21] . Given the close relationship between poly(A) processing and transcription termination of most Pol II transcribed proteincoding genes, it is not clear how multiple functional poly(A) sites within the trypanosome PTU can be transcribed without resulting in premature termination. While little is known regarding the mechanism of Pol II termination in kinetoplastids, two chromatin factors, base J and histone H3 variant (H3.V), have recently been shown to be involved. Base J is a modified DNA base found in kinetoplastids where a glucose moiety, linked via oxygen to the thymine base, resides in the major groove of DNA (reviewed in [22] ). In T. brucei and Leishmania major, J and H3.V are enriched at sites involved in Pol II termination [23] [24] [25] [26] . This includes sites within polycistronic gene clusters where (premature) termination silences downstream genes. Loss of J or H3.V leads to read-through transcription [24] [25] [26] [27] [28] [29] . At PTU internal termination sites this leads to increased expression of the downstream genes [24, 27, 28] . This epigenetic regulation of termination is thought to allow developmentally regulated expression of specific transcriptionally silent genes [24, 27, 28] . In T. brucei, this includes the expression of variant surface glycoprotein (VSG) genes involved in antigenic variation during bloodstream infections [24, 27] .

It is currently unclear how base J and H3.V are involved in Pol II termination, since very little is understood regarding the mechanism of Pol II termination in kinetoplastids. Equally unclear is what regulates the specific localization of these epigenetic marks, including base J, in the genome. Base J is synthesized in a two-step pathway in which a thymidine hydroxylase (TH), JBP1 or JBP2, hydroxylates thymidine residues at specific positions in DNA to form hydroxymethyluracil (hmU) [30] , followed by the transfer of glucose to hmU by the glucosyltransferase, JGT [31, 32] (reviewed in [22, 33] ). The TH enzymes, JBP1 and JBP2, contain a TH domain at the N-terminus [30, [34] [35] [36] [37] . JBP1 has a J-DNA binding domain in the C-terminal half of the protein that is able to bind synthetic J-DNA substrates in vitro and bind chromatin in a J-dependent manner in vivo [38] [39] [40] [41] . The ability of JBP1 to bind J-DNA is thought to play a role in J propagation and maintenance. JBP2 does not bind the modified base directly, but is able to bind chromatin in a base J independent manner, presumably via the C-terminal SWI2/SNF2 domain [34, 38] . The JGT lacks DNA sequence specificity, and can convert hmU to base J in vivo regardless of where it is present [42] [43] [44] . This and other analyses of J synthesis indicate that JBP1 and JBP2 are the key regulatory enzymes of J synthesis.

In this study, we identify a new J-binding protein (called JBP3) in kinetoplastids, which is present in a complex containing PP1, Wdr82 and a putative orthologue of PNUTS. To characterize the role of this (PJW/PP1) complex in transcription termination, we investigated the consequence of mutants using RNAi in T. brucei. Ablation of JBP3, Wdr82 or PNUTS in T. brucei causes read-through transcription at termination sites. As we previously demonstrated following the removal of base J and H3.V, these defects include transcription read-through at termination sites within Pol II transcribed gene arrays and the silent Pol I transcribed VSG expression sites, leading to de-repression of genes involved in parasite pathogenesis. Furthermore, ablation of JBP3, Wdr82 or PNUTS results in expression of genes upstream of Pol II transcription start sites. Presumably this represents a previously unappreciated role of termination of antisense transcription from gene promoters in trypanosomes. Overall these findings provide a first look at mechanisms involved in Pol II transcription termination in kinetoplastids and a direct link between base J and termination.

Base J is synthesized in a two-step pathway involving the hydroxylation of specific thymidine residues in the genome by a thymidine hydroxylase, JBP1 or JBP2, followed by the addition of the glucose moiety by the glucosyltransferase, JGT. In this study, we chose to identify the proteins that co-purify with JBP1, JBP2 and JGT, in order to understand the regulation of J-biosynthesis. Leishmania tarentolae was chosen as an experimental system, because it provides an easily grown source of high densities of parasites that synthesize base J. The proteins were cloned into the pSNSAP1 vector, carrying a C-terminal tag composed of a protein A domain, the TEV protease cleavage site, and the streptavidin binding peptide [45] . Separation of the final eluate from the JGT tagged cell line by SDS-PAGE and staining of the proteins by silver staining, revealed a dominant protein band of~120 kDa, representing the JGT-strep fusion protein (JGT-S) and several co-purified protein bands that are not visible in the control purification from untagged wildtype cells (Fig 1A) . In contrast, and unexpectedly, no co-purified proteins were identified from the JBP1-S or JBP2-S purification either by SDS-PAGE or mass spectrometry analysis. Therefore, we continued the analysis on the JGT purification. Gel based separations followed by in-gel digestion and liquid chromatography-tandem mass spectrometry (LC-MS/MS) of the JGT-S purification revealed a total of five proteins with at least 20 PSMs that were enriched at least 40 fold compared to the negative control purification (Table 1) . A complete list of proteins identified by MS is shown in S1 Table. As expected, JGT-S was recovered in the eluate. The four additional potential JGT associated proteins were discovered with (hypothetical) molecular masses of 74, 42 and 29 kDa. One of these JGT-interacting proteins (42 kDa; LtaP15.0230) was protein phosphatase 1 (PP1), which contains a PP1 catalytic domain (Fig 1B) . The other 42-kDa protein (LtaP32.3990) contains three WD repeat domains and has been identified as a homologue of Wdr82/Swd2 (human/yeast). The remaining two JGT interacting proteins had not been previously characterized. We named the 74-kDa protein JBP3 (LtaP36.0380) because it has a domain with homology to the base J DNA binding domain of JBP1 (Fig 1B and S1 Fig) and we demonstrate its ability to bind J-DNA (see below). The 29-kDa JGT interacting protein was named PNUTS (LtaP33.1440) since it contains a conserved RVxF PP1 interactive domain (Fig 1C and S2 Fig) within an apparent intrinsically disordered region of the protein and is a part of a complex similar to the PTW/PP1 complex in humans (where JBP3 may represent a functional homolog of Tox4). PP1 interactive proteins such as PNUTS are highly disordered in their unbound state and fall in a group of intrinsically disordered proteins (IDPs) [4] [5] [6] . This intrinsic flexibility is important for the formation of extensive interactions with PP1 [6, 46, 47] . A bioinformatics analysis using the DISOPRED3 program [48, 49] , which scores for the occurrence of disorder-inducing amino acids, predicts a majority of TbPNUTS is disordered (S2 Fig). Similarly, the Compositional profiler [50] shows PNUTS is enriched in major disorder-promoting residues and depleted in major order-promoting residues. This inherent disorder may explain why LtPNUTS and TbPNUTS migrate slower in SDS-PAGE than predicted (see Fig 1A and S3A Fig). To confirm the complex, we subsequently performed tandem affinity purifications with tagged JBP3 and Wdr82 followed by shotgun proteomics of the soluble fraction. Reciprocal purification of JBP3 and Wdr82 resulted in the identification of JGT, PNUTS, PP1, JBP3 and Wdr82 with at least 10-fold enrichment of each component compared to WT control purification (S2 Table) . These subsequent purifications and shotgun MS analyses included a replicate of the tagged JGT pull-down. Interestingly, this JGT purification resulted in all components of the complex except PP1 (S2 Table) , suggesting PP1 is the least stable component. These data indicate that [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] and TbPNUTS (residues 139-164), where x is any amino acid and Ø represents a hydrophobic amino acid. Critical residues in the RVxF motif are underlined. (D) Models for the PJW/PP1 complex in Leishmania and T. brucei. The models are based on the human PTW/PP1 complex where PNUTS acts as scaffold and the DNA binding protein and Wdr82 bind to the N-and C-terminus, respectively, and PP1 binds via the RVxF PP1 interaction motif (indicated by the line). PP1 is presumably not stably associated with the complex in T. brucei (see discussion) and is therefore labeled as the PJW complex. (E) Purification of recombinant LtJBP3. His-tagged rJBP3 expressed in E. coli was purified by metal affinity and size exclusion chromatography and analyzed by SDS-PAGE/Coomassie staining. The major copurified protein is the E. coli molecular chaperone GroEL. The migration of protein molecular mass standards (in kDa) is shown on the left. (F) Gel shift assays for modified and unmodified DNA substrates interacting with JBP3. 0.3 pmol radiolabeled J-DNA (J) was incubated with 0, 0.2, 0.38, 0.57 and 1 pmol of JBP3 and 0.3 pmol radiolabeled unmodified DNA (T) was incubated with 1 pmol of JBP3.

https://doi.org/10.1371/journal.pgen.1008390.g001 PNUTS-PP1 regulates transcription termination in T. brucei in Leishmania JGT associates with a protein complex composed of PNUTS, PP1, JBP3 and Wdr82 similar to the PTW/PP1 complex in humans, shown to be involved in transcription termination ( Fig 1D) . JBP3 may be a functional homologue of the human Tox4 DNA binding protein (see below). Based on this similarity, we now refer to this complex as PJW/PP1 in Leishmania.

We noticed that a region of the LtaP36.0380 protein (amino acids 101-277) has sequence similarity with the J-binding domain of JBP1 (S1A and S1B Fig). A structural model of the Lt JBP1 JBD based on the X-ray crystallography-based structure (PDB ID 2xseA) is shown in S1C Fig. As expected from the sequence similarities between the kinetoplastid JBD for JBP1 and the putative JBP3, the 3D models generated by the comparative modeling program I-TASSER [51, 52] were of high quality, with TM-score value of 0.727. The sequence identity and similar domain composition to the JBD of JBP1 supported our contention that the LtaP36.0380 protein might be a J-binding protein, subsequently named JBP3. JBP3, similar to JBP1, appears to be conserved only in kinetoplastids (S4 Fig). We have previously developed a rapid isolation procedure for His-tagged recombinant JBP1 produced in E. coli, and gel shift assays to characterize J-DNA binding activities [40, 41] . We utilized a similar approach to investigate the specific interaction of JBP3 and J-modified DNA (J-DNA). To determine the ability of JBP3 to bind J-DNA, we used the gel shift assay to investigate the binding of recombinant JBP3 ( Fig 1E) to J-DNA duplex (VSG-1J) that has a single centrally located J modification compared with the same duplex without any base J (VSG-T). The J-DNA substrate was incubated with increasing amounts of JBP3 protein, and the complex was analyzed on native gels. The results of the gel shift assay in Fig 1F, show that the amount of free J-DNA decreases with increasing concentrations of JBP3, concomitant with formation of the JBP3/J-DNA complex. In contrast, incubation of the unmodified DNA substrate with the highest concentration of JBP3 resulted in no visible complex. Therefore, JBP3 is a J-DNA binding protein.

To study the function of the PJW/PP1 complex in vivo, we switched to T. brucei due to the benefits of forward and reverse genetics available in this system. However, while all other components are easily identified (see Table 1 ), no homologue of the L. tarentolae PP1 gene is present in the T. brucei genome [53, 54] . To characterize the complex in T. brucei, and identify the L. tarentolae proteins were identified by mass spectroscopy as described in Materials and Methods. Listed are proteins that were enriched at least 40-fold compared to the negative control purification of wildtype extract. The complete list of purified proteins is provided as S1 TbPP1 protein component, we used a TAP tagging approach with a PTP epitope tag [55] to identify TbPNUTS-interacting proteins. We generated a clonal procyclic form (PCF) T. brucei cell line expressing a C-terminal PTP-tagged TbPNUTS protein from its endogenous locus. This cell line was used for the TAP procedure. Briefly, a crude protein extract was first purified by IgG affinity chromatography, and the TEV protease was used to cleave off the Protein A portion of the PTP tag. Subsequently, the TEV protease eluate underwent anti-Protein C affinity purification, and the final purified products were eluted with EGTA. The concentrated proteins were then trypsin-digested and analyzed by shotgun LC-MS/MS. As a control, we analyzed the eluate of a comparable purification of extract from wild-type T. brucei. If we subtract all proteins identified at 1% FDR in the WT control and only consider proteins with a score over 100, we identify seven proteins that include TbPNUTS, Wdr82 and JBP3 (S3 Table) . The other four remaining proteins are commonly identified proteins in immunopurification / MS experiments, including tubulin, heat shock protein and elongation factor 1 [56] . However, neither JGT, nor a protein phosphatase, were identified as part of the TbPNUTS complex. In contrast to Leishmania, base J synthesis is developmentally regulated in T. brucei, where synthesis is lost upon differentiation to PCF due to the downregulation of JBP1 and JBP2 [22, 38, 44, 57] . While JGT has been shown to be part of the nuclear proteome in PCF T. brucei [58] and significant level of JGT activity is present in PCF, as shown by the presence of J in parasites grown in the presence of the J precursor hmU [38, 44] , JGT mRNA is downregu-lated~2-fold between BSF and PCF [59] . Therefore, to fully characterize the association of JGT with the TbPNUTS complex, we need to utilize BSF T. brucei. Co-immunoprecipitation studies were performed to assess the PNUTS-containing complex in BSF T. brucei. Representative components of the putative PP1 complex were analyzed by western blot following immunoprecipitation to assess the authenticity of protein interactions identified by mass spectrometry. We generated BSF T. brucei cell lines expressing PTP-tagged versions of PNUTS, JBP3, Wdr82 and CPSF73 (negative control) (S3A Fig) along with HAtagged versions of JGT, Wdr82 and JBP3 (Fig 2A and S3B Fig) . PTP-PNUTS immunoprecipitation recovers HA-Wdr82 and HA-JBP3, but not the La negative control (Fig 2A) . Similarly, PTP-JBP3 immunoprecipitation recovers HA-Wdr82 (S3B Fig). In contrast, no detectable HA-Wdr82 or HA-JGT is recovered in PTP-CPSF73 immunoprecipitates (Fig 2A and To further characterize the PNUTS complex, extracts recovered from BSF T. brucei cells expressing epitope-tagged components were analyzed by western blot following sucrose fractionation. Analysis of T. brucei cells expressing PTP-PNUTS and HA-JBP3 or HA-Wdr82 indicate Wdr82, JBP3 and PNUTS co-migrate at <200kDa. These results indicate that JBP3, Wdr82 and PNUTS form a stable complex of <200kDa. Summation of the predicted size of the three complex components (~180 kDa) agrees with the observed mass of the PJW complex and suggests a 1:1 stoichiometry for the subunits. Taken together, data from co-immunoprecipitation, sucrose gradient analysis, and identification of PNUTS-, Wdr82-, and JBP3-associated proteins by mass spectrometry indicate that Wdr82, JBP3 and PNUTS comprise a stable PJW complex in kinetoplastids ( Fig 1D) . In contrast to the complex in Leishmania, the stable purified T. brucei complex lacks both JGT and PP1.

LtPP1 is found to be stably associated in the PJW/PP1 complex, possibly through its putative PP1-interacting protein, PNUTS. PNUTS is present in both the T. brucei and Leishmania complexes and share a conserved RVxF motif ( Fig 1C) . Moreover, a similar human PTW/PP1 complex suggests that PP1 confers the complex phosphatase activity critical for its regulation in Pol II termination [2, 3, 60, 61] . The apparent lack of PP1 in the T. brucei complex is therefore surprising. To identify the TbPP1 component, we directly tested two of the eight PP1 genes in T. brucei for interactions with the complex by co-IP. PP1-1 has the highest sequence homology to the PP1 involved in termination in yeast and humans [54] . PP1-7 has been identified in the nucleus of T. brucei [58] . However, neither of these PP1 proteins associates with TbPNUTS in co-IP experiments (S3C Fig) . Another possible explanation for the lack of PP1 in the PNUTS purification, or Co-IP, is that PP1 is not stably associated with the complex in T. brucei. This is consistent with one of four replicates of PJW/PP1 purification that resulted in all components of the complex except PP1 (Table 1) , suggesting PP1 is the least stable component of the Leishmania complex.

If PNUTS, Wdr82 and JBP3 interact with each other, one would predict that RNAi against the individual components would give the same phenotype. We therefore analyzed the role of the PJW/PP1 complex in transcription termination in BSF T. brucei. We have previously shown that base J and H3.V are present at termination sites within a PTU where loss of either epigenetic mark results in read-through transcription and increased expression of downstream genes [24, 27, 28] . Because Pol II elongation and gene expression is inhibited prior to the end PNUTS-PP1 regulates transcription termination in T. brucei of these PTUs, we refer to this as PTU internal termination. For example, base J and H3.V are involved in terminating transcription upstream of the last two genes (VSG; Tb927.5.3990 and Hypothetical protein; Tb927.5.4000) in a PTU on chromosome 5 (Fig 3, top) where deletion of H3.V or inhibition of base J synthesis leads to read-through transcription [24, 27] . The resulting read-through RNAs become observable in a manner similar to those seen in other systems when termination mechanisms become incapacitated by experimental manipulation. The presence of an open reading frame downstream of the termination site allows an additional measure of read-through where nascent RNA is processed to stable capped and polyadenylated Analysis of Pol II termination. Above; Schematic representation of a termination site on chromosome 5 where Pol II has been shown to terminate prior to the last two genes (Tb927.5.3990 and Tb927.5.4000) in the PTU. The dashed arrow indicates readthrough transcription past the TTS that is regulated by base J and H3.V. Solid lines below indicate regions (1 and 2) analyzed by RT-PCR. (C) RT-PCR analysis of nascent RNA. cDNA was synthesized using random hexamers and PCR was performed using the appropriate reverse primer for each region (1 and 2) plus an identical forward primer. A minus-RT control (-RT) is included. Quantitation (fold upregulation) of the PCR products for region 2 of the indicated gels using Gel Doc Quantity One software is provided below. (D) RT-qPCR analysis of genes numbered according to the ORF map above. Transcripts were normalized against Asf1 (which represents a PTU internal Pol II transcribed single copy gene as a negative control), and are plotted as the average and standard deviation of three biological replicates analyzed in triplicate. P values were calculated using Student's t test. �� , p value � 0.01; ��� , p value � 0.001. RT-PCR products for genes 3990 and 4000 were confirmed by cloning and sequencing of multiple clones.

https://doi.org/10.1371/journal.pgen.1008390.g003 PNUTS-PP1 regulates transcription termination in T. brucei mRNA species. As such, the loss of either epigenetic mark in T. brucei leads to generation of nascent RNA extending beyond the termination site and expression of the two downstream genes [24, 27] . To investigate the physiological function of the PJW complex, we analyzed inducible RNAi ablation of Wdr82, JBP3, and PNUTS in BSF T. brucei. As shown in Fig 3, induction of RNAi against Wdr82, JBP3 and PNUTS and ablation of mRNA levels from 30-60% leads to reduced parasite growth, indicating that the proteins are important for normal cell proliferation in BSF T. brucei. In the case of PNUTS, ablation of mRNA levels to 50% leads to~90% reduction at the protein level (S5 Fig). We also detect evidence of read-through transcription at the representative PTU internal termination site on chromosome 5 upon ablation of the three factors. RT-PCR using oligos flanking the termination site (see diagram, Fig 3A) detects increased RNA upon ablation of PNUTS, Wdr82 and JBP3 ( Fig 3C) . As a control a separate RT-PCR utilized the same 5' primer and a 3' primer immediately upstream of the termination site. We have previously shown that an RNA species spanning the termination site is indicative of read-through transcription and is only detected following the loss of base J or H3. V, due to continued transcription elongation at termination sites [24, 27, 28] . Consistent with read-through transcription, both genes downstream of the termination site are significantly de-repressed upon the ablation of the three components of the PJW complex, in contrast to genes upstream ( Fig 3D) . In contrast, no significant termination defects are detected upon ablation of a negative control, acidocalcisome VA a protein (Fig 3C and 3D ). VA a has been shown to be an essential gene in T. brucei and ablation results in growth defects similar to those seen in PJW complex [62] (Fig 3A) . Therefore, the read-through defects measured in the PNUTS, Wdr82 and JBP3 mutants are presumably not the result of indirect effects of dying cells. These results suggest that PNUTS, Wdr82 and JBP3, possibly functioning in the PJW complex, are essential for Pol II termination.

To further explore the role of the PJW complex in the regulation of termination and whether the complex functions similarly across the genome, we performed stranded mRNAseq to compare the expression profiles of PNUTS RNAi cells with and without tetracyclin induction. This led to the detection of 709 mRNAs that are increased at least 3-fold upon ablation of PNUTS (P adj <0.05) (S6A Fig and S4 Table) . In contrast, no mRNAs are downregulated. Of the 3-fold upregulated genes, a majority represent VSGs, ESAGs and Retrotransposon Hot Spot proteins (RHS) that are repressed in bloodstream T. brucei and are localized at the end of Pol II transcribed PTUs located within the chromosomes or in subtelomeres. The location of the genes with >3-fold upregulation was mapped ( Fig 4A and S7 Fig) . Interestingly, these genes were closely located at regions flanking PTUs or subtelomeric regions, suggesting a role of PNUTS in genome-wide transcription, specifically at transcription initiation and termination sites, as well as subtelomeres. As evident in the genome-wide view (S7 Fig) , a significant fraction of the upregulated genes represent the silent subtelomeric VSG and RHS gene arrays while the majority of the regulated genes adjacent to PTU flanking regions in the chromosome core are annotated as hypothetical proteins of unknown function (Fig 4A and S7 Fig  and S4 Table) . The PTU flanking regions represent transcription termination sites (TTS) and transcription start sites (TSS) within the so-called convergent strand switch region (cSSR) and divergent strand switch region (dSSR), respectively. We have previously shown that base J is localized within the cSSRs and dSSRs flanking PTUs and involved in Pol II termination. Therefore, we wanted to first confirm the association with J at termination sites with changes in gene expression we observed in the PNUTS mutant. If we analyze genes within 1kb of base J genome-wide, we see a correlation with upregulation of gene expression versus a similar analysis of the same number of randomly selected genes across the genome (S6B Fig). In contrast, the 4 genes upregulated in the VA a mutant lack any association with base J. The upregulated genes in the PNUTS mutant located at the 3' end of PTUs, including the VSG gene (Tb927.5.3990) analyzed in Fig 3, map to regions downstream of base J and H3.V. The RNAseq results confirm our initial RT-PCR analysis of nascent and steady-state RNA indicating a role of PNUTS, Wdr82 and JBP3 regulating termination and expression of downstream genes (Fig 3) . In these studies, we are measuring gene expression levels at day 2 of the RNAi to avoid potential secondary effects due to cell growth defects. While many of these changes reflect increases from silent or extremely low initial levels of RNA, and thus represent small total increases in mRNA levels, the changes are reproducible, significant and therefore, we believe to be biologically relevant. Consistent with our analysis of the J/H3.V mutants, we observe In cSSR, the poly A sites of the final gene in the PTU (indicated by the transcriptome) are marked by dotted line (0 kb). In dSSR, the dotted line (0 kb) indicates the 5' end of the first gene in the PTU (according to the transcriptome). Thus, the TSS is located further upstream, within the dSSR. Numbers refer to distance from SSR (kb). Below: Metagene profile of total sense and antisense RNA-seq signal over the SSRs and 5kb upstream and downstream regions into the PTUs. Fold changes comparing transcript levels between day 0 and day 2 following induction of RNAi were calculated and plotted over the indicated regions genome-wide. Red and blue lines indicate RNAs from the top and bottom strand, respectively, as indicated on the diagram above. PNUTS, PNUTS RNAi; VA a, RNAi VA a. (D) Above: Diagram of a HT site. Boxes are the genes and arrows indicate direction of transcription as in C. The center of the H2A.Z peak is marked by a dotted line. Numbers refer to distance from center of H2A.Z peak (kb). Below: Metagene profile of the fold changes in RNA-seq reads over the HT sites and 10 kb upstream and downstream of the H2A.Z peak, as described in C. similar increases in the PNUTS mutant in the expression of genes downstream of PTU internal termination sites, which we previously demonstrated is caused by a defect in Pol II transcription termination resulting in readthrough transcription [24, 27, 28] .

A remaining question was whether PNUTS also regulated termination at the 3'-end of gene clusters resulting in transcription into the SSR. Our previous RNA-seq analyses in T. brucei indicated that H3.V/J regulated termination at the 3'-end of PTUs, where mutants resulted in Pol II elongation into the dual strand transcription regions at cSSRs [24, 27] . To visualize the differences between WT and TbPNUTS mutant, fold changes between -Tet and +Tet induction of RNAi were plotted (S8 Fig). While sense transcription remained largely unaffected throughout all 11 megabase chromosomes following the loss of PNUTS, significant fold increases of antisense transcription were observed near transcription borders of PTUs, including downstream of normal transcription termination sites at the 3'-end. To take a closer look at the increased level of transcripts at these sites, and determine whether they are due to transcriptional readthrough, forward and reverse reads mapping to 5kb flanking and within cSSRs were counted and RPKM values were generated. As shown in Fig 4B, cSSRs are computationally defined regions where coding strands switch based on the transcriptome. To see the difference between WT and PNUTS mutant, changes in transcript abundance upon PNUTS and VA a ablation were plotted. A metaplot summarizing the readthrough defect for all cSSRs is shown in Fig 4B. We observed an increase in RNA extending into the cSSR in the PNUTS mutant, and no change in the VA a mutant control. Boxplots comparing the median RPKM values for SSRs also indicated that cSSRs were upregulated in the PNUTS mutant and the differences were statistically significant (S9 Fig) . All together these results indicate that the loss of PNUTS, JBP3, and Wdr82 result in defects in transcription termination at the 3'-end of PTUs as we described in the H3.V/J mutant.

Based on the initial RT-PCR analysis of the internal termination site in PNUTS, JBP3 and Wdr82 knockdowns (Fig 3) , we expected RNA-seq to illustrate defects in termination at 3'end of PTUs genome-wide in the PNUTS mutant. However, we also detected accumulation of transcripts at dSSRs, illustrated by the genome-wide map of transcript levels (S8 Fig) and the metaplot of dSSRs (Fig 4C and S9 Fig) . At these divergent TSSs, total RNA levels detected in the region upstream of the start site are increased in the PNUTS mutant when forward and reverse reads were analyzed separately, suggesting that transcription initiates upstream of its start site upon the loss of PNUTS. In some cases, this leads to expression of genes present in the dSSR that are silent in WT cells. This explains the mapping of genes upregulated in the PNUTS mutant to 5'-end PTUs (Fig 4A) . A specific example, shown in Fig 5A, includes a dSSR on chromosome 10 where a gene (Tb927.10.8340) located between the two divergent TSSs (mapped by tri-phosphate RNA sequencing of WT cells) is affected by the loss of PNUTS. Specific upregulation of this gene 4-to 13-fold following the loss of PNUTS, JBP3 and Wdr82 versus genes in the adjacent PTU is confirmed by RT-qPCR ( Fig 5B) . Another example confirmed by RT-qPCR is shown in S10 Fig where the gene (Tb927.10.6430) located upstream of the TSS in WT cells is specifically upregulated 4-to 7.5-fold by the loss of PNUTS, JBP3 and Wdr82.

Many promoters for RNA Pol II are bidirectional in organisms from yeast to human [63] [64] [65] [66] [67] . Unidirectional transcription resulting in productive mRNAs is typically ensured since antisense transcription is susceptible to early termination linked to rapid degradation. Mapping TSSs using tri-phosphate RNA-seq has previously suggested bi-directional activity, with strong strand bias, of Pol II initiation sites in T. brucei [68, 69] . The RNA-seq data presented here is consistent with bi-directional activity of TSS, early termination and stimulation of divergent antisense transcription in the PNUTS knockdown leading to the activation of genes within the dSSR (Figs 4C and 5A and S10 Fig) . As shown in Fig 5A and 5B , the silent 8340 gene within a dSSR on chromosome 10 is transcribed upon the loss of TbPNUTS. To see whether there is a correlation between the initiation of the antisense transcript with sense RT-qPCR analysis was performed for the indicated genes as described in Material and Methods. Transcripts were normalized against Asf1 mRNA, and are plotted as the average and standard deviation of three replicates. P values were calculated using Student's t test. �� , p value � 0.01; ��� , p value � 0.001. RT-PCR products for gene 8340 were confirmed by cloning and sequencing of multiple clones. (C) Mapping of the 5' end of the antisense transcript by nested RT-PCR on strand-specific cDNA. PCR utilized constant 3' primer (indicated by black arrow-head) with varying 5' primers indicated in red (primers 1-4). cDNA levels utilized in the PCR reactions were normalized against strand-specific Asf1 mRNA. (D) Strand-specific RT-qPCR analysis of antisense nascent transcript. cDNA was generated using various strand specific 3' primers and RNA from-and + Tet. qPCR was then done using internal primers to amplify the region indicated by black bar. Transcript levels were normalized against strandspecific Asf1 mRNA. Error bars indicate standard deviation from at least three experiments.

https://doi.org/10.1371/journal.pgen.1008390.g005 PNUTS-PP1 regulates transcription termination in T. brucei mRNA coding strand TSS, we analyzed the 5'-end of the nascent antisense transcript at this dSSR using RT-PCR (Fig 5C) . The significant drop of inability of a 5' primer to amplify the cDNA corresponding to the nascent 'antisense' transcript indicates the 5' end of the PNUTS regulated transcript is adjacent to the TSS for the sense transcript on the opposing strand. Using various 3' primers for generating cDNA, antisense transcription is attenuated in uninduced cells as shown in the decline in cDNA with increasing length of the transcript (Fig 5D) . In the absence of PNUTS the antisense transcription fails to terminate, as seen in the maintained levels of cDNA for all primers tested in + Tet. The increasing effects of the loss of PNUTS on level of transcript with increasing length of the transcript supports the idea of PNUTS regulating early termination/elongation of the antisense transcript. Taken together the results suggest that the PJW complex (PNUTS, JBP3 and Wdr82) regulates premature termination of antisense transcription from bi-directional TSS and silencing of gene expression in T. brucei.

Another possibility is that increased transcription within dSSRs is due to transcription initiating upstream of the divergent TSSs in the absence of PNUTS. To address this possibility, we examined transcription at Head-to-Head (HT) boundaries of PTUs. HT sites are defined where transcription of one PTU terminates and transcription of another PTU on the same DNA strand initiates. In contrast to TSSs at dSSRs, HT sites contain a termination site for the upstream PTU, indicated by H3.V and H4.V, and a TSS marked by individual peaks of histone variants, such as H2A.Z, and histone modifications [25, 68, 69] . Similar to the metaplot analyses of dSSRs and cSSRs, we detected accumulation of transcripts at HT sites upon the ablation of PNUTS (Fig 4D) . At these non-divergent TSSs, the region upstream of the start site produced more transcripts in the PNUTS mutant indicating readthrough transcription, as expected, or possible initiating events further upstream. More interestingly, antisense transcripts are also increased at HT sites. The lack of a divergent TSS at these sites, indicated by the presence of single peak of H2A.Z, supports the conclusion that the antisense transcripts are the result of regulated bi-directional transcription activity.

Noncoding transcription, via defects in transcription termination, influences eukaryotic replication initiation. Transcription through origins located at 5'-and 3'-ends of Pol II transcription units leads to replication defects via dissociation of the prereplication complex (pre-ORCs) or sliding of MCM helicases [70] [71] [72] [73] [74] . Of the 40 early firing origins that have been mapped in T. brucei, 36 are upstream of TSS [75] . Analysis of L. major has indicated that replication initiation sites occur at the genomic locations where Pol II stalls or terminates, including sites precisely downstream of base J [76] . Therefore, increased transcription at PTU flanking regions in the absence of TbPNUTS may cause DNA replication defects. To see whether pervasive transcription has any effect on T. brucei replication, we analyzed whether TbPNUTS is required for proper cell-cycle progression. In T. brucei, DNA replication and segregation of kinetoplastid DNA (K) in the single mitochondrion precede those of nuclear DNA (N), so cells at different stages can be distinguished by their N and K configurations. 

In addition to Pol II termination sites distributed throughout the T. brucei genome, H3.V and base J localize within the~14 Pol I transcribed polycistronic units located at the telomeric ends and involved in antigenic variation (so called bloodstream form VSG expression sites, BESs) ( Fig 6A) [22, [77] [78] [79] . Monoallelic expression of a VSG ES leads to the expression of a single VSG on the surface of the parasite, a key aspect of the strategy bloodstream form trypanosomes use to evade the host immune system. We have previously shown that the loss of H3.V and J leads to increased expression of VSG genes from silent telomeric BESs [24] . This effect is presumably due to the role of these epigenetic marks in attenuating transcription elongation Pol I within the silent VSG BESs, thereby preventing the transcription of silent VSGs. Differential gene expression analysis of RNA-seq reads mapped to the 427 genome indicates that the loss of PNUTS leads to increased VSG expression from silent BESs (S2 Table) . In addition to the BES, depletion of TbPNUTS results in de-repression of VSGs from the silent telomeric metacyclic ES (MES), which are transcribed monocistronically by Pol I. A few of these VSG gene expression changes have been confirmed by RT-qPCR (S12 Fig). To further explore the global function of TbPNUTS in VSG expression control, we mapped the RNA-seq reads to the VSGnome [80] . The VSGnome allows the analysis of VSG genes, such as those on minichromosomes, that were not included in the new T. brucei 427 genome assembly. As shown in Fig  6, this analysis confirms the de-repression of Pol I transcribed VSG at BES and MES upon TbPNUTS depletion. On the other hand minichromosomal (MC) VSGs lacking a promoter [81] were not significantly affected; only 2 of 41 were upregulated (P adj <0.1). These data indicate that transcription of (silent) telomeric VSGs in the PNUTS mutant is strongly dependent on the presence of a Pol I promoter. Interestingly, comparing the expression level of MES VSGs that are adjacent to the promoter with BES VSGs over 10 kb downstream (Fig 6A and  S12 Fig) suggested that the level of derepression is a function of distance from the Pol I promoter. As previously mentioned, the majority of VSG genes upregulated in the PNUTS mutant are chromosomal internal VSGs (S4 Table and S7 Fig). In the VSGnome, these (unknown) VSGs ( Fig 6B) were thought to be primarily located at subtelomeric arrays, but their exact positioning in the genome was not known [80] . These VSGs have now been mapped to the silent subtelomeric arrays assembled in the new 427 genome [79] and, as shown in S7 Fig and  Fig 6B , a significant fraction are upregulated upon the loss of PNUTS. Overall, these data indicate that TbPNUTS (PJW complex) regulates transcriptional silencing of telomeric Pol I transcribed telomeric VSG PTUs and Pol II transcription of VSGs within genome internal PTUs in T. brucei.

The derepression of ESAG 6 and 7 genes adjacent to the BES promoter along with the VSG 40 kb downstream (S4 Table) , suggests that PJW may function throughout the telomeric PTU. To examine this more closely, we analyzed the RNA-seq reads mapping to the 14 telomeric BES sequences [82] . RNA-seq reads mapping to BES were counted in 200bp windows with a 100bp steps. Read counts were converted into reads per million (RPM) and compared between +/-Tet to estimate log2 fold change and plotted in Fig 6C and S13 Fig. The location of BES promoters is indicated by an arrow. The transcription of the active BES1 was not affected by the loss of TbPNUTS. However, when the remaining 13 silent BESs are analyzed we see derepression of ESAGs as well as the terminal VSG. In some cases, it seems that derepression extends 10-20 kb from the promoter to express ESAG 6 and 7, with no significant effect on the remaining genes (ESAGs) within the silent BES, and derepression of the VSG at the 3' end. In other cases there is selective upregulation of the VSG gene, including the pseudo VSGs that are present upstream of some telomeric VSGs. The apparent selective VSG upregulation may be due to the combined effect of the low level of transcription of the derepressed BES and enhanced stability of VSG mRNAs. For example, the increased VSG mRNA half-life (4.5 hrs) compared with ESAG 6 and 7 (1.8-2.8 hrs) [83, 84] . Transcripts of genes located PNUTS-PP1 regulates transcription termination in T. brucei approximately 5 kb upstream of the VSG have also been shown to be selectively rapidly degraded [85] . We also noticed that increased levels of RNA close to the promoter are significantly higher when there is an additional BES promoter upstream for ESAG10. In fact, the most significant gene expression changes are in BESs that have dual promoters (S4 Table) . These data would suggest derepression of telomere-proximal VSG genes after PNUTS depletion is due to transcriptional activation of silent Pol I promoters. However, these results are also consistent with increased Pol I elongation along the BES. Repression of silent ESs is mediated at least in part by the inhibition of Pol I elongation within the BES preventing the production of VSG mRNA from the silent BESs [86] [87] [88] . Similar to its inhibition of Pol II transcription elongation at termination sites at the 5' and 3' end of PTUs genome-wide, PNUTS may also function at telomeric regions to attenuate Pol I transcription elongation within the silent ESs. The data here suggest PJW controls VSG silencing at BESs by regulating Pol I elongation (termination) and/or regulating access of the polymerase to silent promoter regions.

Current dogma in the field is that most, if not all, Pol II transcribed gene regulation is at the posttranscriptional level in kinetoplastids [89] . This is primarily based on the polycistronic arrangement of genes and identification of posttranscriptional regulons under the control of RNA binding protein regulated mRNA stability. Our previous studies on base J/H3.V function revealed regulated transcription termination at the 3'-end of PTUs as a novel mechanism of gene silencing in kinetoplastids. The studies described here solidify this concept and open up another possible regulatory gene expression mechanism via early termination at the 5'-end of PTUs. These studies also provide the first direct mechanistic link between base J and transcription termination in kinetoplastids by the identification of a multi-subunit protein complex involved in termination that binds base J.

Here we identify a new base J binding protein, JBP3, and show that it is part of a PP1-PNUTS-Wdr82 containing protein module. We named this module the PJW/PP1 module based upon the mammalian PTW/PP1 module involved in transcription termination [9] . Mutation of its components, JBP3, Wdr82 and PNUTS, gives similar phenotypes in T. brucei, validating PJW is indeed a functional module. Our data strongly suggests that the kinetoplastid PJW/PP1 complex we identified here is reminiscent of human PTW/PP1 where PNUTS is the kinetoplastid functional homologue of human PNUTS and JBP3 the homologue of Tox4. However, while we demonstrate LtPP1 is a component of the Leishmania complex, the purified module in T. brucei lacks a PP1 homologue. PP1 is the only catalytic component of the human PTW/PP1 complex and dephosporylation by PP1 is directly involved in regulating Pol II termination [2, 3, 60, 61] . Therefore, the apparent lack of PP1 association with the PNUTS, Wdr82 and JBP3 complex involved in transcription termination in T. brucei is surprising. MS analysis of the TbPNUTS complex is the result of a single purification from 12 L of PC T. brucei cells. The finding that one of the four PJW/PP1 complex purifications in Leishmania also failed to pulldown PP1 suggests PP1 may be the least stable component of the complex in T. brucei. Once we increase the yield of TbPNUTS purification, and reduce the volume of cells needed, we can perform multiple pulldown/MS analyses to address this possibility. The specific loss of PP1 and JBP3 in one of the LtJGT pulldowns, may also reflect an effect of the tandem affinity tag on LtJGT associations. Thus, future experiments should examine T. brucei complex purification using a different affinity tag or using TbJBP3 or TbWdr82 PTP-fusion proteins. We propose that PP1 is a functional component of the PJW/PP1 complex in both Leishmania and T. brucei, but association with TbPNUTS is less stable. However, further work needs to be done to determine whether PP1 is a component of the PTW complex and involved in termination in T. brucei. Therefore, we refer to the PJW complex when addressing functional analysis of the complex in T. brucei and, based on our studies in Leishmania, the PJW/ PP1 complex when discussing the potential role of the complex in termination in kinetoplastids.

The association of JGT, the glucosyltransferase involved in base J synthesis, with the PJW/ PP1 complex in Leishmania is also unexpected. Further work is needed to clarify the function of JGT in the Leishmania PJW/PP1 complex, for example, in regulating base J synthesis. Reciprocal tagging of JBP3 and Wdr82 pulls down the same PJW/PP1 components in Leishmania, including JGT. This helps confirm the nature of the PJW/PP1 complex and suggests that the tag itself does not significantly influence the protein interactions. As mentioned above, we have not directly addressed the possibility that the lack of PP1 in the TbPNUTS-PTP purification could be due to the PTP-tag. However, the lack of JGT in the T. brucei PJW complex is likely not due to the PTP-tag since co-IP analysis using tagged versions of PNUTS, JBP3 or Wdr82 fail to pull down JGT in BSF parasites. Furthermore, ablation of PNUTS, JBP3 and Wdr82 in BSF T. brucei has no significant effect on level of base J in the genome (S14 Fig). Not only does this support the lack of JGT of in the T. brucei PJW complex, but also addresses the possibility that termination defects we measure in the PJW mutants are due to the loss of base J.

We and others have previously shown that base J and H3.V co-localize at Pol II termination sites [23, 25] and are involved in transcription termination in T. brucei and L. major, where loss of base J and H3.V leads to readthrough transcription at the 3'-end of PTUs [24, [26] [27] [28] [29] . For 'premature' termination sites within PTUs, read-through transcription resulting from the loss of base J and/or H3.V leads to transcription of silent genes downstream of the termination site [24, 27, 28] . We now show that depletion of components of the PJW complex in T. brucei leads to similar defects in Pol II termination at the 3'-end of PTUs, including the de-repression of downstream genes. We also uncover additional defects at the 5'-end of PTUs suggesting regulated early termination of antisense transcription from bi-directional transcription start sites. Genes in T. brucei have been shown to exhibit PTU positioning bias, where genes located near the 3'-end of a PTU exhibit relative increase in relative abundance of mRNAs upon heat shock [90] . To control for the stress of dying cells, we analyzed the ablation of the VA protein that results in a similar growth defect as the ablation of PJW complex components. In contrast to the 709 genes upregulated in the PNUTS RNAi, VA ablation led to the increased expression of only four genes and no significant changes in read-through at 3'-ends of PTUs (Fig 4B) . However, a small increase in RNA from dSSRs was evident in the VA RNAi ( Fig 4C) . Interestingly, this significantly smaller effect corresponds with a slightly reduced growth defect in the VA RNAi compared to the PNUTS RNAi. We are unable to rule out the possibility that some of the increased RNAs we detect in these studies are in response to stress. For example, some treatments that inhibit T. brucei growth (such as cold shock and mild acid) have resulted in increased expression of procyclin [91, 92] , a gene that is upregulated in both the PNUTS and VA mutant. However, the apparent increase in RNAs from dSSRs of the VA control RNAi does not lead to any gene expression changes, since none of the four upregulated genes localize to this region of the genome. As we previously described following the loss of base J and H3.V in T. brucei and Leishmania, detection of increased nascent, unprocessed RNA by RT-PCR at the termination site supports the conclusion that the loss of PJW complex function leads to Pol II read-through transcription, rather than some Pol I or Pol II (re-) initiation event. The lack of identifiable Pol I promoter at these regions, and the low level of resulting transcripts where RNA Pol I transcription is known to be 5-10 times more active than transcription by RNA Pol II [93] , supports this conclusion. Although we cannot exclude a potential role of the PJW complex in the regulation of RNA processing, the increase in both unprocessed (represented by nascent RNA spanning the termination site and RNA-seq reads corresponding to intergenic regions) and processed RNAs (mRNAs) strongly suggests the complex regulates RNA abundance at the level of transcription and that the defects we observe are not simply due to an alteration of RNA processing. We conclude that the gene expression changes we see in the PNUTS mutant reflect the role of the PJW complex in Pol II transcription termination in T. brucei.

Here, we propose a model where divergent transcription at the 5'-end and readthrough transcription at the 3'-end of gene arrays is affected by the PJW/PP1 complex in kinetoplastids (model in Fig 7) . According to this model, the PJW/PP1 complex is recruited to termination sites, at least partially, due to base J-JBP3 interactions. H3.V localized at 3'-end termination According to the model, bi-directional transcription initiates at each TSS, but only the Pol II transcribing the 'sense strand' fully elongates and generates productive poly (A) mRNAs. Pol II terminates the 3'-end of the PTU marked with base J (ball and stick) and H3.V. In T. brucei, the PJW complex is recruited to the termination site via JBP3 recognition of J-DNA and somehow controls termination. In Leishmania, regulation of termination by the PJW/PP1 complex presumably involves dephosphorylation of the CTD of Pol II (as discussed in the text). While we propose a similar mechanism in T. brucei, these studies did not identify PP1 as a stable component of the PJW complex. The PJW complex is also recruited to the dSSR upstream of the TSS, leading to premature termination of antisense transcription. These short transcripts may be additionally targeted for degradation. Impaired termination, following mutation of the PJW complex, leads to readthrough transcription at the 3'-and 5'-end of PTUs. Genes located downstream of these termination sites, at the 3'-and 5'-end (upstream of the TSS) of a PTU, can generate stabilized polyA mRNAs and be expressed. PNUTS-PP1 regulates transcription termination in T. brucei sites may play an additional role in complex localization since WDR5, a homolog of Wdr82, binds to the N-terminal tail of histone H3 [94] . Wdr82 is required for recruitment of the APT termination complex containing PNUTS-PP1 to snoRNA termination sites in yeast [95] . Wdr82 may also play a role in 5'-end termination site recognition since it binds to RNA Pol II CTD phosphorylated at Ser5 in yeast and humans [10, 96] .

Similar to the mammalian complex, we propose PNUTS is a scaffolding protein for the entire PJW/PP1 complex and regulates PP1 function via the PP1 binding RVxF motif. Only three substrates have been identified for PNUTS/PP1: the Pol II elongation factor Spt5, the CTD of Pol II and MYC [2, 60, 97, 98] . MYC dephosphorylation by PP1 regulates chromatin binding and stability. PP1 dephosphorylation of Spt5 and Pol II has been directly implicated in regulating Pol II termination in other eukaryotes. Therefore, we propose that regulated phosphorylation of Spt5 and Pol II by the PJW/PP1 complex is critical for transcription termination in trypanosomatids. TbSpt5 has recently been shown to be associated with Pol II [99] and is phosphorylated at a single Ser residue [100] . The CTD of Pol II in trypanosomatids is unique in that it does not contain the heptad or other repetitive motifs that are conserved from yeast to humans [101] . However, the CTD was shown to be essential for Pol II transcription in T. brucei [102, 103] and 17 phosphorylated sites have been identified within the CTD [100] . Studies also suggest that CTD phosphorylation is required for Pol II association with trypanosome chromatin [104] . It is necessary to determine the role of the PJW/PP1 complex in the phosphorylation status of Spt5 and Pol II in trypanosomatids and whether phosphorylation of a non-canonical Pol II CTD is involved in transcriptional regulation in divergent eukaryotes thought to lack significant regulation.

An unexpected finding of this work was that, in the absence of PNUTS, Wdr82 or JBP3, genes located upstream of TSS are also now expressed, resulting from transcription between diverging PTUs. Mammalian and yeast promoters frequently give rise to transcription in both the sense and divergent antisense directions [63] [64] [65] [66] [67] , giving rise to a (productive) sense transcript and a corresponding upstream noncoding RNA (ncRNA) [105] [106] [107] [108] . Unidirectional transcription is typically ensured since the ncRNAs are susceptible to early termination linked to rapid degradation [106, [109] [110] [111] . Early termination of divergent transcription at 5' ends of mammalian genes occurs by similar mechanisms as termination at 3' ends. In addition to regulating termination at the 3' end of genes, Wdr82 and PNUTS have also been shown to be involved in enforcing early transcription termination at bi-directional promoters [17] . The role of PNUTS is thought to reflect the essential nature of PNUTS/PP1 since differential phosphorylation of the CTD of Pol II has been proposed to regulate the directionality of transcription at bi-directional promoters. Specifically, Tyr1 and Ser1 hyperphosphorylation of Pol II have been shown to be associated with antisense divergent transcription at mammalian promoters [112] [113] [114] . Spt5 regulation of Pol II elongation is involved in control of divergent antisense transcription as well as readthrough transcription at 3' end of genes in yeast [2, 115] , representing an additional target of PNUTS/PP1 regulation of transcription at both ends of genes. Similar to mammalian promoters where transcription is divergent and initiation is over a broad genomic region, previous studies have suggested that Pol II transcription initiation sites are intrinsically bi-directional in T. brucei [68, 69] . We found that loss of PNUTS, Wdr82 and JBP3 can have a major effect on divergent 'antisense' transcription, presumably reflecting the role of PJW in regulating termination of antisense transcription from bi-directional promoters in T. brucei (Fig 7) . Interestingly, decreased levels of base J in Leishmania led to detection of antisense RNAs corresponding to similar regions within divergent PTUs [26] . A possible explanation for antisense transcription at divergent TSSs is increased chromatin accessibility in the SSR, resulting in alternative TSS usage further upstream. However, the detection of significant antisense transcription at non-divergent TSSs (HT sites) upon the loss of PNUTS strongly supports the involvement of the PJW in regulating early termination of antisense transcription in T. brucei. Additional work is required to unambiguously confirm that the increased nascent antisense RNA, and corresponding mRNAs, we detect from regions upstream of TSSs are in fact a result of bi-directional transcription activity.

Loss of J and H3.V in T. brucei resulted in similar readthrough defects at the 3'-end of PTUs (including telomeric BESs) and gene expression as seen here in PJW mutants, but without cell growth effects. We concluded that readthrough transcription at 3'-end and corresponding gene expression changes in the J/H3.V mutant are not lethal to the parasite. PJW mutants we analyze here lead to additional effects on transcription and gene de-repression at 5'-ends and decreased cell growth. The ability of the PJW complex to bind base J is consistent with its function at both ends of PTU's. However, it is unclear why the effects in the J mutant are limited to the 3'-end and whether specific function of the complex at the 5'-end can explain its essential nature. While J is present at both 5' and 3' PTU flanking regions and involved in transcription, it is apparently not the dominant mark since H3.V had more significant effects on 3'-end transcription and gene expression [24] . H3.V is limited to termination sites at the 3'ends of PTUs and not localized at TSSs in T. brucei. A significant role of H3.V in PJW complex function, as discussed above, would explain why H3.V and PJW mutants share termination defects at the 3'-end. Similarly, additional specific chromatin factors may be involved in recruitment of PJW complex to TSSs independent of base J, such as Pol II-Wdr82 interactions, as mentioned above, and modified and variant histones such as H2A.Z and H3K4Me3 (Fig 7) . These points may explain why PJW mutants lead to defects at both 5'-and 3'-ends and the J/ H3.V mutants are limited to the 3'-end. We propose that the essential nature of the PJW complex is due to regulated expression of ncRNAs and/or mRNAs at the 5'-end of PTUs or additional unknown functions of JBP3, Wdr82 and PNUTS in T. brucei.

The possibility that the essential nature of the complex is due to regulation of transcriptionreplication conflicts at TSSs was directly addressed. We expected the pervasive transcription phenotype in the PJW/PP1 mutant to negatively impact DNA replication and explain the reduced cell growth. Replication origins tend to localize after TTS in yeast and upstream of promoters in humans, presumably to minimize transcription-replication conflicts [73, 116, 117] . The induction of transcription through origins, via defects in transcription termination at TTSs as well as at TSSs, leads to replication defects via dissociation or sliding of the pre-ORCs and MCMs [70] [71] [72] [73] [74] and changes in chromatin structure [118] . Furthermore, loss of origin function activates readthrough transcription in mammals and yeast. Clearly there is a relationship between transcription and DNA replication in eukaryotes. This relationship appears to exist in T. brucei as well, since origins flank the PTUs at TSS and TTS, and TbORC1 and TbMCM-BP deletions led to similar defects in transcription at PTU flanks as we illustrate here [75, 119] . In the recent analysis of the TbMCM-BP mutant [119] , in addition to its role in termination at 3' ends, the antisense RNA at H-T sites suggested a role of TbMCM-BP in determining the direction of transcription, what we refer to as bi-directional activity, while at dSSRs they concluded it was due to alternative initiation events. Regardless of the mechanism involved, increased transcription upstream of TSSs in the PJW/PP1 mutants would presumably result in defects in replication. However, the TbPNUTS mutant does not indicate any alteration in cell cycle or DNA replication. Alternatively, the level of transcription induction at origins is too low at day two of the RNAi for any significant effects on replication. Further work is needed to explore the effects of increased transcription on ORC function in trypanosomes.

Functional interaction between replication and transcription machineries was further suggested by derepression of Pol I transcribed silent BESs and MESs in the TbORC1 and TbMCM-BP mutants [75, 120] . How these operate mechanistically at ESs, regulating Pol I elongation via chromatin changes along silent BESs or BES promoter activity, remains unclear. In fact, whether VSG monoallelic expression control of BESs takes place at the initiation or elongation level is still debated. Low levels of transcripts from silent ESs upon the knockdown of factors such as ORC1, MCM-BP and PNUTS cannot resolve this issue, since the data are compatible with both models. While derepression of the BES in the PNUTS mutant would suggest the PJW complex has a direct effect on the activity of silent BES promoters, the proposed role of the complex in regulating Pol II elongation/termination at 5' and 3' end of PTUs genome-wide would suggest a similar role in regulating Pol I elongation. If so, presumably Pol I is regulated in a different manner, since there is no evidence for phosphorylation of Pol I. However, the Spt5 substrate for PNUTS-PP1 has been shown to bind and regulate Pol I transcription in mammalian cells [121, 122] . Further studies are necessary to understand how PJW complex regulation of Pol II transcription at 5' and 3' ends of PTUs genome-wide is related to silencing of the specialized telomeric Pol I PTUs. For example, derepression of BES promoters may be a functional consequence of non-coding RNA transcription generated by pervasive Pol II transcription in the PNUTS mutant. We have previously shown that readthrough transcription at 3' ends of PTUs lead to significant levels of siRNAs in T. brucei [24] . Regardless of the mechanism involved, the complex is required not only for repression of telomeric and subtelomeric VSGs but also VSGs scattered within the chromosome at 3' ends of PTUs. Depletion of PNUTS also increased the levels of procyclin and PAG RNAs, which are transcribed from a Pol I promoter that is repressed in BSF T. brucei. This transcription unit is located at the end of Pol II transcribed PTU and increased transcription in PNUTS mutant may be due to readthrough transcription as in the J/H3.V mutant [24] . Thus, the PJW complex is required for repression of life-cycle specific genes transcribed by Pol I in the mammalian infectious form of T. brucei. We have therefore uncovered a possible functional link between transcriptional termination and Pol II-and Pol I-mediated gene silencing in T. brucei.

Bloodstream form T. b. brucei Lister 427 (MiTat 1.2) or "single marker cells (SMC)", expressing T7 RNA polymerase and the Tet repressor [123] were used in these studies and cultured in HMI-9 medium. Transfections were performed using the Amaxa electroporation system (Human T Cell Nucleofactor Kit, program X-001). Where appropriate, the following drug concentrations were used: 2 μg/ml G418, 5 μg/ml Hygromycin, 2.5 μg/ml Phleomycin, 2 μg/ml Tetracycline. Procyclic form T. b. brucei TREU667 and promastigote form L. tarantolae were cultured in SDM79 medium. Transfections were performed using the BioRad GenePulser II (2 pulses at 1.4 kV / 25 μF) in 0.4 cm cuvettes with 1 x 10 8 cells (L. tarantolae) or 3 x 10 7 cells (PCF T. b. brucei) in 0.5 ml cytomix (2 mM EGTA, 120 mM KCl, 0.15 mM CaCl 2 , 10 mM KP i pH = 7.6, 25 mM HEPES pH = 7.6, 5 mM MgCl 2 , 0.5% glucose, 100 μg/ml BSA, 1 mM Hypoxanthine). Where appropriate, the following drug concentrations were used: 25 μg/ml G418 (PCF T. b. brucei) and 50 μg/ml G418, 10 μg/ml Puromycin (L. tarantolae).

For conditional PNUTS, JBP3 and Wdr82 silencing experiments in T. brucei a part of the ORF was integrated into the BamH I site of the p2T7-177 vector [124] . Sce-I linearized p2T7-177 constructs were transfected into BF SMC for targeted integration into the 177bp repeat locus. RNAi was induced with 2 μg/ml Tetracycline and cell growth was monitored daily in triplicate. Primers sequences used are available upon request.

For generation of the dual (Protein A and Streptavidin Binding Protein) tagged constructs in L. tarantolae, the coding regions of LtJGT, LtJBP3, and LtWdr82 lacking stop codons were amplified and cloned into the BamHI and XbaI sites of pSNSAP1 [45] . The resulting constructs are referred to as JGT-SA, JBP3-SA and Wdr82-SA. In Lt. cells ectopically expressing the JGT-S, we performed an additional KO for a single JGT WT allele. For PTP tagging in T. brucei, the 3' end of T. brucei genes were cloned in the Apa I and Not I sites of the pC-PTP-Neo vector [55] , resulting in dual (Protein A and Protein C) tagged proteins. Linearization of the constructs was performed using a unique restriction site within the 3' end of the cloned gene. All final constructs were sequenced prior to electroporation. Tagging the 3'-end of the TbPNUTS, TbJBP3 and TbWdr82 with 3x HA tag was performed using a PCR based approach with the pMOTag4H construct [125] . Primers sequences and construct information used are available upon request.

To quantify the genomic J levels, DNA was isolated and utilized in the anti-J DNA immunoblot assay as described previously [126] . Briefly, serially diluted genomic DNA was blotted to nitrocellulose, followed by incubation with anti-J antisera. Bound antibodies were detected by a secondary goat anti-rabbit antibody conjugated to HRP and visualized by ECL. The membrane was stripped and hybridized with a probe for the beta-tubulin gene to correct for DNA loading. Pull-down experiments using PTP-tagged PNUTS, JBP3 and Wdr82 were performed using the first purification step of the TAP protocol on whole cell extracts from 2 x 10 8 cells in 200 μl buffer II. After the IgG Sepharose incubation, beads where washed in PA-150 buffer and boiled in Laemmli buffer. For Western analysis, 5 x 10 6 cell equivalents of each fraction (input, unbound and bound) were analyzed on 10% PAA / SDS gels and sequentially probed with anti-HA antibodies (Sigma, 3F10, 1:3000), anti-Protein A antibodies (Sigma, P3775, 1:5000) and anti-La antibodies (a gift from C. Tschudi, 1:500).

For the initial JGT-S purification, purified proteins from JGT tagged cells and negative control cells were separated by SDS-PAGE gel and visualized by silver staining. Each lane of the silverstained SDS-PAGE gel was divided into 5 large fractions before being cut down further into roughly 2mm x 2mm pieces and stored in separate tubes in preparation for digestion. Each fraction of gel was de-stained before undergoing denaturation in 10 mM dithiothreitol at 56˚C for one hour. Denatured proteins were then alkylated by 55mM iodoacetamide for 45 minutes in the dark with intermittent vortexing. Sequencing grade trypsin (Promega, V5111) was then utilized to digest proteins overnight at 37˚C with gentle vortexing. Peptides were extracted from each fraction separately by incubating gel pieces with a gradient of increasing concentrations of acetonitrile. All extracted peptides were concentrated, reconstituted in 0.1% formic acid in 5% acetonitrile, and passed through a 0.2um bio-inert membrane tube (PALL, ODM02C35) to remove any residual particulate. Each fraction was injected independently into an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific) interfaced with an UltiMate 3000 RSLCnano HPLC system (Thermo Scientific). Peptides were resolved on an Acclai PepMap RSLC C18 analytical column (75um ID x 15cm; 2μm particle size) at a flow rate of 200nL/min using a gradient of increasing buffer B (80% acetonitrile in 0.1% formic acid) over 180 minutes. Data dependent acquisition was carried out using the Orbitrap mass analyzer collecting full scans every three seconds (300-2000 m/z range at 60,000 resolution). The most intense ions that met mono-isotopic precursor selection requirements were selected, isolated, and fragmented using 38% collision-induced dissociation (CID). Every precursor selected for ms/ms analysis was added to the dynamic exclusion list and precursors selected twice within 10 seconds were excluded for the following 20 seconds. Fragment ions were detected using the ion trap to increase the duty cycle and achieve more ms/ms scans per experiment. Raw data for each fraction was searched separately against the appropriate databases: GeneDB Leishmania tarentolae and T. brucei 927 database, using the Sequest search algorithm in Proteome Discoverer 2.2 (Thermo Fisher Scientific). Search parameters were set to allow for two missed tryptic cleavages, 20 ppm mass tolerance for precursor ions, and 0.3 Daltons mass tolerance for fragment ions. A fixed modification of carbamidomethylation on cysteine residues and a variable modification of oxidation on methionine residues were enabled to accurately match fragment ions. A fixed value peptide spectral match (PSM) validation node was used to validate each PSM at a maximum delta Cn of 0.05. True negative PSMs (decoy PSMs) were generated by searching the same raw data utilizing a decoy database, containing reversed protein sequences from the target protein database, and the exact same search parameters. All surviving PSMs were imported into ProteoIQ (v2.7, Premier Biosoft) for final validation. ProteoIQ used validated PSMs and decoy PSMs to generate protein identifications with a maximum false discovery rate of 1%. Each protein present at 1% FDR was also required to have at least two unique peptides and 10 PSMs to be reported. For shotgun proteomics, eluted proteins were digested in-solution as previously described [127] and then analyzed by LC-MS/MS as described above.

To analyze the putative J-DNA binding domain of JBP3, the representative region of LtJBP3 was threaded through the JBP1 JBD PDB structure to search for similar secondary structural folds using the comparative modeling program I-TASSER (iterative threading assembly refinement algorithm). (http://zhang.bioinformatics.ku.edu/I-TASSER) The initial 3D models generated by I-TASSER were of high quality, with a C-score of -0.76 and a TM-score of 0.62 (within the correct topology range of I-TASSER). TM-score is defined to assess the topological similarity of the two protein structures independent of size. A TM-score >0.5 corresponds approximately to two structures of the similar topology. The top predicted model was then aligned using TM-align from I-TASSER with the Lt JBD model 2xseA giving a TM-score of 0.727, RMSD of 1.46, and Cov score of 0.774.

The L. tarentolae JBP3 gene was amplified from genomic DNA by PCR with primers containing 5' Nde I and 3' BamH I restriction sites. PCR fragments were digested with Nde I and BamH I and subcloned into the pet16b expression vector. Plasmids were transformed into Escherichia coli (BL21 DE3), and bacteria were cultured in defined autoinduction media to allow growing cultures to high densities and protein expression without induction, as previously described [128] . Briefly, LB media is supplemented with KH 2 PO 4 , Na 2 HPO 4 and 0.05% Glucose, 0.2% Lactose and 0.6% Glycerol. Bacteria were cultured in this media in the presence of 100μg/ml ampicillin at 37˚C for 2 h and then shifted to 18˚C for 24 h. Cells were lysed and JBP3 purified by metal affinity as previously described for JBP1 [40, 41] . The affinity-purified JBP3 was concentrated to 0.5 ml in a Centricon-100 apparatus and loaded onto a Sephadex S-200 (Amersham Biosciences 16/60) column equilibrated with buffer A (50 mM Hepes, pH 7.0, 500 mM NaCl, 1 mM DTT). The fraction containing JBP3 was concentrated to 200 ul by Centricon-100. Protein purity was analyzed by SDS-PAGE stained with Coomassie Brilliant Blue, and protein concentration was determined using BSA standards.

Electrophoretic Mobility Shift Assays were carried out essentially as described previously for JBP1 and VSG-1J DNA substrate [40, 41] , with few changes. The binding reaction mixture (20 ul) contained 35 mM Hepes-NaOH, pH 7.9, 1mM EDTA, 1 mM DTT, 50 mM KCL, 5 mM MgCl, 10 ug BSA and 15 fmol radiolabeled DNA substrate containing a single modified J base (VSG-1J) or no modified base, and indicated JBP3 protein amounts. The reactions were incubated for 15 min at room temperature and analyzed on a 4.5% nondenaturing polyacrylamide gel at 4˚C. After drying, the gels were exposed to film.

For the sedimentation analysis of the PJW/PP1 complex, extracts were made from the BF T. brucei cell lines expressing PNUTS-PTP and JBP3-HA or Wdr82-HA and loaded onto 10 ml, 5-50% sucrose gradients. Samples were centrifuged at 38000 rpm for 18 hours using a SW41 Ti rotor (Beckman). The gradient was fractionated from top to bottom in twenty aliquots of 500ul each. Proteins from each fraction were enriched by methanol: chloroform protein precipitation and resuspended in SDS loading buffer for electrophoresis.

Total RNA was isolated with Tripure Isolation Reagent (Roche). cDNA was generated from 0.5-2 μg Turbo DNase (ThermoFisher) treated total RNA with Superscript III (ThermoFisher) according to the manufacturer's instructions with either random hexamers, oligo dT primers or strand specific oligonucleotides. Strand specific RT reactions were performed with the strand specific oligonucleotide and an antisense Asf I oligonucleotide. Equal amounts of cDNA were used in PCR reactions with Ready Go Taq Polymerase (Promega). A minus-RT control was used to ensure no contaminating genomic DNA was amplified. Primer sequences used in the analysis are available upon request.

Total RNA was isolated and Turbo TM DNase treated as described above. Quantification of Superscript TM III generated cDNA was performed using an iCycler with an iQ5 multicolor real-time PCR detection system (Bio-Rad). Triplicate cDNA's were analyzed and normalized to Asf I cDNA. qPCR oligonucleotide primer combos were designed using Integrated DNA Technologies software. cDNA reactions were diluted 10 fold and 5 μl was analyzed. A 15 μl reaction mixture contained 4.5 pmol sense and antisense primer, 7.5 μl 2X iQ SYBR green super mix (Bio-Rad Laboratories). Standard curves were prepared for each gene using 5-fold dilutions of a known quantity (100 ng/μl) of WT gDNA. The quantities were calculated using iQ5 optical detection system software. Primers sequences used are available upon request.

For mRNA-seq, total RNA was isolated from T. brucei RNAi cultures grown in presence or absence of tetracycline for two days using TriPure. Six mRNA-seq libraries were constructed for PNUTS RNA (triplicate samples for plus and minus tetracyclin) and four libraries for VAa RNA (duplicate samples) were constructed using Illumina TruSeq Stranded RNA LT Kit following the manufacturer's instructions with limited modifications. The starting quantity of total RNA was adjusted to 1.3 μg, and all volumes were reduced to a third of the described quantity. High throughput sequencing was performed at the Georgia Genomics and Bioinformatics Core (GGBC) on a NextSeq500 (Illumina).

Raw reads from mRNA-seq were first trimmed using fastp with default settings (v0.19.5; [129] ). Remaining reads were locally aligned to the recently published long-read T. brucei Lister 427 2018 version 9.0 genome assembly (downloaded from [80] ) and the Lister 427 BES sequences [82] , using Bowtie2 version 2.3.4.1 [130] . With non-default settings (sensitive local) and further processed with Samtools version 1.6 [131] . To ensure proper read placement, alignments with multiple low-quality hits and mapping quality (MQ) scores less than 10 were removed. For each sample, HTSeq (v0.9.1) was used to count reads for each reference transcript annotation, followed by normalization/variance stabilization using DESeq2 (v1.18.1). Differential gene expression was conducted using DESeq2 by comparing TbPNUTs RNAi samples with and without tetracyclin in triplicate (log2 fold change and differential expression test statistics can be found in S3 Table) . To express the transcripts levels for individual mRNA encoding genes as shown in S3 Table, we estimated transcript abundance as transcripts per million (TPM) by first normalizing the number of reads by kilobase of transcript, and then scaling each transcript per sample such that the sum of all transcript abundances within a sample was equal to 1 million. Due to incomplete gene annotation of the BES in the new T. brucei Lister 427 2018 genome assembly, gene expression changes for BES were determined by aligning reads to the Lister 427 BES sequences (S4 Table) . To compare tetracyclin-treatment fold changes for specific strands genome-wide, we counted reads from each strand in 200bp bins with a 100bp step. Mapping of differentially upregulated genes in a genome-wide context was determined by highlighting genes upregulated >3-fold in red for all 11 megabase chromosomes. Fold changes between Tet-untreated and Tet-treated PNUTS RNAi were plotted for all 11 megabase chromosomes as well. Tag counts in 200bp bins (100bp steps) were used to estimate correlations among samples (correlation coefficients among replicates were >0. 99) .

To analyze transcription defects at 3'-and 5'-end of PTUs, reads mapping to TTS (cSSRs) and TSS (dSSRs) were counted and reads per kilobase per million mapped reads (RPKM) values were generated. Similar to what we have previously done [24, 27, 28] , lists of cSSRs and dSSRs were generated computationally as defined regions where coding strands switch based on the transcriptome as well as TSS mapped via triphosphate RNA sequencing. Head-to-Tail (HT) sites were defined were one PTU terminates (H3.V) and another PTU on the same strand initiates. The TSS at HT sites were further defined by a single peak of H2A.Z and the lack of an annotated gene on the antisense strand, distinguishing it from a TSS at a dSSR. Several SSRs located at subtelomeres were not included due to ambiguous nature of gene organization. SSRs and the 5-kb flanking regions were analyzed with DeepTools (v3.2.1) using 100bp bins flanking SSRs and dividing each SSR into 50 equally sized bins. Violin plots were generated using the R package vioplot, with the median and interquartile range illustrated by white circles and black boxes, respectively. Genes were considered adjacent to base J if the gene, according to the T. brucei Lister 427 annotation, was within 1-kb either upstream or downstream base J peaks. J IP-seq data shown here are from previously published work [23, 25] .

To examine VSG expression, trimmed reads were bowtie-aligned to the VSGnome (retrieved from http://tryps.rockefeller.edu) [80] and differential expression was analyzed using the DEseq2 identically to the analysis of the entire genome.

The cell cycle profile of the PNUTS RNAi cell line over time following induction was determined by staining cells with DAPI and cataloguing the nucleus/kinetoplastid (N/K) configurations of~300 intact cells/time-point. Cells were harvested, 2 x washed in PBS, allowed to settle on glass slides and air dried for 10 min at RT. Cells were then fixed in -20˚C methanol for 10 min and mounted with ProLong Gold antifade reagent with DAPI (Molecular Probes, P36935). Flow cytometry was carried out by washing the cells 2 times in PBS and resuspended at 1 x 10 7 / ml. Ice cold 100% ethanol was slowly added while shaking to a final concentration of 25% ethanol. Cells were then washed 2 times and resuspended in 0.5 ml PBS / 1% BSA / 1% Tween 20 with 5 μg / ml Propidium Iodide. Analysis was performed using a Cyan cytometer (DAKO). [132] using jackhammer [133] until convergence with an e-value cut-off of 0.01 for sequence and 0.03 for hits. Full length protein sequences from representative species were chosen and aligned using hmmalign [133] and alignancer, trimmed using trimAL [134] with the automated1 flag, and a phylogenetic tree was made using raxml [135] (options -f a -x 12345 -p 12345 -N autoMRE -m PROTGAMMAJTTF). Matching sequences were found exclusively in species within the Kinetoplastida class. The sequences separated into two distinct groups with high bootstrap support. We were able to find JBP-3 family members in all kinetoplastid genomes where JBP-1 is found, apart from Strigomonas culicis, and Perkinselia sp. which only contain JBP-1 family members. (14 BESs) . Fold changes comparing plus and minus Tetracyclin were plotted over each BES as described in Fig 6C. (TIF) S14 Fig. PJW does not affect the synthesis of base J. DNA was isolated from the indicated T. brucei cell lines for anti-J dot blot analysis. +Tet indicates samples from day 2 RNAi induction. Samples were 2-fold serially diluted. The same blots were hybridized with a radioactive tubulin probe to control for DNA loading. (TIF) S1 Table. JGT purification and gel based MS analysis in L. tarentolae. LtJGT was purified, fractionated on SDS-PAGE and proteins identified by mass spectrometry as in Table 1 . Complete list of proteins identified in the JGT and the WT negative control purifications is shown. Proteins that were enriched at least 40-fold (based on PSMs) compared to the negative control purification are highlighted. (DOCX) S2 Table. Mass spectrometric identification of JGT, JBP3 and Wdr82 purification products. LtJGT, LtJBP3 and LtWdr82 proteins were purified and proteins in the soluble fraction identified by shotgun proteomics. Proteins that were enriched at least 10-fold (based on PSMs) compared to the negative control purification are highlighted. Pep: Peptides; Cov: Coverage. (DOCX) S3 Table. PNUTS purification and MS analysis in T. brucei. TbPNUTS was purified and proteins in the soluble fraction identified by shotgun proteomics. List of PNUTS-PTP co-purified proteins, identified at 1% FDR by LC-MS/MS, with 10 or more PSMs are shown. Proteins that are enriched at least 40-fold compared to the negative control purification of wildtype extract and a score of at least 100 are highlighted. Included are proteins that are common contaminants of previous tandem affinity purifications (i.e. tubulin and elongation factor 1-alpha). Protein annotation and accession numbers are from the T. brucei 927 database at www.TriTrypDB.org. 

The transcription cycle in eukaryotes: from productive initiation to RNA polymerase II recycling

Elongation/Termination Factor Exchange Mediated by PP1 Phosphatase Orchestrates Transcription Termination

RNA polymerase II termination involves C-terminal-domain tyrosine dephosphorylation by CPF subunit Glc7

The extended PP1 toolkit: designed to create specificity

Detailed structural characterization of unbound protein phosphatase 1 inhibitors

Spinophilin directs protein phosphatase 1 specificity by blocking substrate binding sites

Subunit structure and regulation of protein phosphatase-1 in rat liver nuclei

Purification and characterisation of p99, a nuclear modulator of protein phosphatase 1 activity

Identification and characterization of a novel human PP1 phosphatase complex

Wdr82 is a C-terminal domain-binding protein that recruits the Setd1A Histone H3-Lys4 methyltransferase complex to transcription start sites of transcribed human genes

A role for SSU72 in balancing RNA polymerase II transcription elongation and termination

Functional interactions between the transcription and mRNA 3' end processing machineries mediated by Ssu72 and Sub1

CPF-associated phosphatase activity opposes condensin-mediated chromosome condensation

Organization and function of APT, a subcomplex of the yeast cleavage and polyadenylation factor involved in the formation of mRNA and small nucleolar RNA 3'-ends

The essential WD repeat protein Swd2 has dual functions in RNA polymerase II transcription termination and lysine 4 methylation of histone H3

The Glc7 phosphatase subunit of the cleavage and polyadenylation factor is essential for transcription termination on snoRNA genes

Transcription of Mammalian cis-Regulatory Elements Is Restrained by Actively Enforced Early Termination

3' processing in protists

Gene expression in Kinetoplastids

Trans-splicing in trypanosomes: machinery and its impact on the parasite transcriptome

Gene expression in Trypanosoma brucei: lessons from high-throughput RNA sequencing

Base J: discovery, biosynthesis, and possible functions

Two thymidine hydroxylases differentially regulate the formation of glucosylated DNA at regions flanking polymerase II polycistronic transcription units throughout the genome of Trypanosoma brucei

Histone H3 Variant Regulates RNA Polymerase II Transcription Termination and Dual Strand Transcription of siRNA Loci in Trypanosoma brucei

Four histone variants mark the boundaries of polycistronic transcription units in Trypanosoma brucei

Glucosylated hydroxymethyluracil, DNA base J, prevents transcriptional readthrough in Leishmania

Regulation of transcription termination by glucosylated hydroxymethyluracil, base J, in Leishmania major and Trypanosoma brucei

Base J represses genes at the end of polycistronic gene clusters in Leishmania major by promoting RNAP II termination

Base J and H3.V Regulate Transcriptional Termination in Trypanosoma brucei

JBP1 and JBP2 Proteins Are Fe2 +/2-Oxoglutarate-dependent Dioxygenases Regulating Hydroxylation of Thymidine Residues in Trypanosome DNA

Identification of the glucosyltransferase that converts hydroxymethyluracil to base J in the trypanosomatid genome

Tb927.10.6900 encodes the glucosyltransferase involved in synthesis of base J in Trypanosoma brucei

Enzymatic Formation of the Hypermodified DNA Base J

JBP1 and JBP2 are two distinct thymidine hydroxylases involved in J biosynthesis in genomic DNA of African trypanosomes

a SWI2/SNF2-like protein, regulates de novo telomeric DNA glycosylation in bloodstream form Trypanosoma brucei

Evidence that J-binding protein 2 is a thymidine hydroxylase catalyzing the first step in the biosynthesis of DNA base

The protein that binds to DNA base J in trypanosomatids has features of a thymidine hydroxylase

Regulation of trypanosome DNA glycosylation by a SWI2/ SNF2-like protein

The structural basis for recognition of base J containing DNA by a novel DNA binding domain in JBP1

Recognition of base J in duplex DNA by J-binding protein

Site-specific interactions of JBP with base and sugar moieties in duplex J-DNA

Base J glucosyltransferase does not regulate the sequence specificity of J synthesis in trypanosomatid telomeric DNA

A method for the efficient and selective identification of 5-hydroxymethyluracil in genomic DNA

Biosynthesis and function of the modified DNA base beta-D-glucosyl-hydroxymethyluracil in Trypanosoma brucei

Isolation of a U-insertion/deletion editing complex from Leishmania tarentolae mitochondria

Structural basis for regulation of protein phosphatase 1 by inhibitor-2

Structural basis of protein phosphatase 1 regulation

DISOPRED3: precise disordered region predictions with annotated proteinbinding activity

The DISOPRED server for the prediction of protein disorder

Composition Profiler: a tool for discovery and visualization of amino acid composition differences

I-TASSER: a unified platform for automated protein structure and function prediction

The I-TASSER Suite: protein structure and function prediction

The TriTryp phosphatome: analysis of the protein phosphatase catalytic domains

Okadaic acid overcomes the blocked cell cycle caused by depleting Cdc2-related kinases in Trypanosoma brucei

Highly efficient tandem affinity purification of trypanosome protein complexes based on a novel epitope combination

The CRAPome: a contaminant repository for affinity purification-mass spectrometry data

2-Oxoglutarate-dependent hydroxylases involved In DNA base J (β-D-Glucopyranosyloxymethyluracil) synthesis

The nuclear proteome of Trypanosoma brucei

TriTrypDB: a functional genomic resource for the Trypanosomatidae

PNUTS/PP1 regulates RNAPII-mediated gene expression and is necessary for developmental growth

Mass spectrometry and biochemical analysis of RNA polymerase II: targeting by protein phosphatase-1

Proteomic analysis of the acidocalcisome, an organelle conserved from bacteria to human cells

Native elongating transcript sequencing reveals human transcriptional activity at nucleotide resolution

Widespread bidirectional promoters are the major source of cryptic transcripts in yeast

Mammalian NET-Seq Reveals Genome-wide Nascent Transcription Coupled to RNA Processing

RNA exosome depletion reveals transcription upstream of active human promoters

Bidirectional promoters generate pervasive transcription in yeast

The transcriptome of the human pathogen Trypanosoma brucei at single-nucleotide resolution

GT-rich promoters can drive RNA pol II transcription and deposition of H2A.Z in African trypanosomes

Post-licensing Specification of Eukaryotic Replication Origins by Facilitated Mcm2-7 Sliding along DNA

Perturbation of the activity of replication origin by meiosis-specific transcription

The requirement of yeast replication origins for pre-replication complex proteins is modulated by transcription

Genome-wide identification of replication origins in yeast by comparative genomics

Transcription interferes with elements important for chromosome maintenance in Saccharomyces cerevisiae

Genome-wide analysis reveals extensive functional interaction between DNA replication initiation and transcription in the genome of Trypanosoma brucei

Transcriptionally Driven DNA Replication Program of the Human Parasite Leishmania major

Antigenic variation in African trypanosomes

Molecular mechanisms underlying the control of antigenic variation in African trypanosomes

Genome organization and DNA accessibility control antigenic variation in trypanosomes

Capturing the variant surface glycoprotein repertoire (the VSGnome) of Trypanosoma brucei Lister 427

The small chromosomes of Trypanosoma brucei involved in antigenic variation are constructed around repetitive palindromes

Telomeric expression sites are highly conserved in Trypanosoma brucei

RNA turnover in Trypanosoma brucei

Studies on the recycling of the transferrin receptor in Trypanosoma brucei using an inducible gene expression system

Selective inhibition of RNA polymerase I transcription as a potential approach to treat African trypanosomiasis

Stimuli of differentiation regulate RNA elongation in the transcription units for the major stage-specific antigens of Trypanosoma brucei

An update on antigenic variation in African trypanosomes

Differential RNA elongation controls the variant surface glycoprotein gene expression sites of Trypanosoma brucei

Post-transcriptional regulation of gene expression in trypanosomes and leishmanias

Genome organization is a major component of gene expression control in response to stress and during the cell division cycle in trypanosomes

Cold shock and regulation of surface protein trafficking convey sensitization to inducers of stage differentiation in Trypanosoma brucei

Mild acid stress as a differentiation trigger in Trypanosoma brucei

The PARP promoter of Trypanosoma brucei is developmentally regulated in a chromosomal context

Histone H3 recognition and presentation by the WDR5 module of the MLL1 complex

Yeast Swd2 is essential because of antagonism between Set1 histone methyltransferase complex and APT (associated with Pta1) termination factor

Targeted recruitment of Set1 histone methylase by elongating Pol II provides a localized mark and memory of recent transcriptional activity

MYC dephosphorylation by the PP1/PNUTS phosphatase complex regulates chromatin binding and protein stability

A Cdk9-PP1 switch regulates the elongation-termination transition of RNA polymerase II

An RNA polymerase II-associated TFIIF-like complex is indispensable for SL RNA gene transcription in Trypanosoma brucei

Global quantitative SILAC phosphoproteomics reveals differential phosphorylation is widespread between the procyclic and bloodstream form lifecycle stages of Trypanosoma brucei

Comparative genomics of cyclin-dependent kinases suggest co-evolution of the RNAP II C-terminal domain and CTD-directed CDKs

An essential domain of an earlydiverged RNA polymerase II functions to accurately decode a primitive chromatin landscape

The non-canonical CTD of RNAP-II is essential for productive RNA synthesis in Trypanosoma brucei

Stress induces changes in the phosphorylation of Trypanosoma cruzi RNA polymerase II, affecting its association with chromatin and RNA processing

Nascent RNA sequencing reveals widespread pausing and divergent initiation at human promoters

RNA polymerase plays both sides: vivid and bidirectional transcription around and upstream of active promoters

Divergent transcription of long noncoding RNA/mRNA gene pairs in embryonic stem cells

Functional consequences of bidirectional promoters

Antisense RNA polymerase II divergent transcripts are P-TEFb dependent and substrates for the RNA exosome

Polyadenylation site-induced decay of upstream transcripts enforces promoter directionality

Transcriptome surveillance by selective termination of noncoding RNA synthesis

Tyrosine phosphorylation of RNA polymerase II CTD is associated with antisense promoter transcription and active enhancers in mammalian cells

RNA Pol II Dynamics Modulate Co-transcriptional Chromatin Modification, CTD Phosphorylation, and Transcriptional Direction

Tyrosine-1 of RNA Polymerase II CTD Controls Global Termination of Gene Transcription in Mammals

Spt5 Plays Vital Roles in the Control of Sense and Antisense Transcription Elongation

Genome-wide mapping of human DNA-replication origins: levels of transcription at ORC1 sites regulate origin selection and replication timing

Selectivity of ORC binding sites and the relation to replication timing, fragile sites, and deletions in cancers

Noncoding transcription influences the replication initiation program through chromatin regulation

Genome-wide function of MCM-BP in Trypanosoma brucei DNA replication and transcription

Trypanosoma brucei Orc1 is essential for nuclear DNA replication and affects both VSG silencing and VSG switching

The transcription elongation factor Spt5 influences transcription by RNA polymerase I positively and negatively

Yeast transcription elongation factor Spt5 associates with RNA polymerase I and RNA polymerase II directly

A tightly regulated inducible expression system for conditional gene knock-outs and dominant-negative genetics in Trypanosoma brucei

Targeting of a tetracycline-inducible expression system to the transcriptionally silent minichromosomes of Trypanosoma brucei

A vector series for rapid PCR-mediated C-terminal in situ tagging of Trypanosoma brucei genes

Localization of the modified base J in telomeric VSG gene expression sites of Trypanosoma brucei

Defining the regulated secreted proteome of rodent adipocytes upon the induction of insulin resistance

Catalytic function and substrate specificity of the papain-like protease domain of nsp3 from the Middle East respiratory syndrome coronavirus

fastp: an ultra-fast all-in-one FASTQ preprocessor

Fast gapped-read alignment with Bowtie 2

The Sequence Alignment/Map format and SAMtools

UniProt: a worldwide hub of protein knowledge

We are grateful to Jessica Lopes da Rosa-Spiegler, David Reynolds, Nicolai Siegel, and Piet Borst for critical reading of the manuscript. We would also like to thank Shulamit Michaeli (Bar-IIan University) for the pSNSAP1 vector; Arthur Günzl (UConn Health) for the pC-PTP-NEO vector; Roberto DoCampo (University of Georgia) for VA a RNAi construct; Mike Cipriano (University of Georgia) for help with homology modeling; Christina L. Ethridge (University of Georgia) for generating the RNA-seq libraries.