RNA Polymerase II CTD phosphatase Rtr1 prevents premature transcription termination

RNA Polymerase II (RNAPII) transcription termination is regulated by the phosphorylation status of the C-terminal domain (CTD). Using disruption-compensation (DisCo) protein-protein interaction network analysis, interaction changes were observed within the termination machinery as a consequence of deletion of the serine 5 RNAPII CTD phosphatase Rtr1. Interactions between RNAPII and the cleavage factor IA (CF1A) subunit Pcf11 were reduced in rtr1Δ, whereas interactions with the CTD and RNA-binding termination factor Nrd1 were increased. These changes could be the result of altered interactions between the termination machinery and/or increased levels of premature termination of RNAPII. Transcriptome analysis in rtr1Δ cells found decreased pervasive transcription and a shift in balance of expression of sense and antisense transcripts. Globally, rtr1Δ leads to decreases in noncoding RNAs that are linked to the Nrd1, Nab3 and Sen1 (NNS)-dependent RNAPII termination pathway. Genome-wide analysis of RNAPII and Nrd1 occupancy suggests that loss of RTR1 leads to increased termination at noncoding genes and increased efficiency of snRNA termination. Additionally, premature termination increases globally at protein-coding genes where NNS is recruited during early elongation. The effects of rtr1Δ on RNA expression levels were erased following deletion of the exosome subunit Rrp6, which works with the NNS complex to rapidly degrade terminated noncoding RNAs. Overall, these data suggest that Rtr1 restricts the NNS-dependent termination pathway in WT cells to prevent premature RNAPII termination of mRNAs and ncRNAs. Additionally, Rtr1 phosphatase activity facilitates low-level elongation of noncoding transcripts that impact the transcriptome through RNAPII interference. AUTHOR SUMMARY Many cellular RNAs including those that encode for proteins are produced by the enzyme RNA Polymerase II. In this work, we have defined a new role for the phosphatase Rtr1 in the regulation of RNA Polymerase II progression from the start of transcription to the 3’ end of the gene where the nascent RNA from protein-coding genes is typically cleaved and polyadenylated. Deletion of the gene that encodes RTR1 leads to changes in the interactions between RNA polymerase II and the termination machinery. Rtr1 loss also causes early termination of RNA Polymerase II at many of its target gene types including protein coding genes and noncoding RNAs. Evidence suggests that the premature termination observed in RTR1 knockout cells occurs through the termination factor and RNA binding protein Nrd1 and its binding partner Nab3. Additionally, many of the prematurely terminated noncoding RNA transcripts are degraded by the Rrp6-containing nuclear exosome, a known component of the Nrd1-Nab3 termination coupled RNA degradation pathway. These findings suggest that Rtr1 normally promotes elongation of RNA Polymerase II transcripts through preventation of Nrd1-directed termination.


INTRODUCTION 18
The termination of transcription by eukaryotic RNA Polymerase II (RNAPII) is 19 tightly coupled with RNA processing including small RNA processing, splicing, and 20 mRNA cleavage and polyadenylation at the 3'-end of protein-coding genes [1]. Recent 21 studies have reaffirmed that transcription termination in eukaryotes is a highly dynamic 22 process that can lead to different gene expression outputs through mechanisms such as 23 alternative polyadenylation site usage and premature transcription termination [2][3][4][5][6][7][8][9]. 1 Transcription termination in yeast has been shown to be regulated through numerous 2 termination factors as well as the phosphorylation status of the C-terminal domain 3 (CTD) of RNAPII, which has the repetitive sequence (Tyr 1 -Ser 2 -Pro 3 -Thr 4 -Ser 5 -Pro 6 -4 Ser 7 )n [10,11]. However, the exact mechanisms that underlie the role of CTD 5 dephosphorylation in the regulation of elongation, termination, and the attenuation of 6 these processes remain unclear. At least four phosphatases are components of the 7 yeast transcription elongation/termination machinery: Rtr1, Ssu72, Glc7, and Fcp1 [12-8 21]. There appears to be extensive interplay between the protein phosphatases and 9 their control of the phosphorylation status of the RNAPII CTD. For instance, serine 5 10 (Ser5) dephosphorylation has been shown to be carried out by both Rtr1 and Ssu72 in 11 both in vivo and in vitro studies [12,[22][23][24][25]. Additionally, Ssu72 dephosphorylation of 12 Ser5 serves as a prerequisite for Ser2 dephosphorylation by Fcp1 [13,15]. However, it 13 remains unclear how temporal dephosphorylation impacts the formation and/or 14 recruitment of RNA processing complexes during transcription and the determination of 15 the termination pathway that will be used. 16 17 One pathway that is heavily influenced by CTD phosphorylation is the Nrd1, 18 Nab3 and Sen1 (NNS) polyadenylation independent transcription termination pathway 19 [26][27][28][29][30][31]. Nrd1 contains a RNAPII CTD interaction domain (CID) that preferentially 20 interacts with a Ser5-P CTD [26]. The NNS complex regulates transcription termination 21 of short non-coding transcripts and transcription elongation of a selection of protein 22 coding genes [28,[32][33][34][35]. Pcf11, a member of the cleavage factor Ia complex (CFIa), 23 also contains a CID that has increased affinity for Ser2-P over Ser5-P modified RNAPII 1 CTD. Pcf11 has been shown to be required for both poly-adenylation dependent 2 termination and NNS termination [26,36,37]. While the cleavage and polyadenylation 3 factor (CPF) complex does not contain any known CID containing proteins, both Ssu72 4 and Glc7 are integral subunits of CPF. Rtt103 a CID-containing protein with specificity 5 to Ser2-P CTD, has been proposed to form a higher order complex with CFIa and CPF 6 to possibly bridge the Rat1 exoribonuclease to the RNAPII CTD to trigger degradation 7 of the cleaved 3' end of the RNA transcription product and hence transcription 8 termination [26,38]. Rat1, Rtt103, and the decapping nuclease Rai1 are sufficient to 9 terminate elongating RNAPII in vitro and have been shown to be required for RNAPII 10 termination in vivo [39][40][41][42]. However, numerous subunits of CFIa and CPF are required 11 for fully efficient RNAPII termination in vivo suggesting that higher-order interactions 12 between the transcription termination machinery, RNAPII, and the RNA are likely 13 required in eukaryotes [43,44]. Additional factors such as Rrp6, a subunit of the nuclear 14 exosome, may also play a role in transcription termination through targeting of certain 15 cellular states of RNAPII such as the backtracked enzyme (previously described as the 16 reverse torpedo model) [43,[45][46][47]. We propose that the extensive control of RNAPII 17 CTD dephosphorylation in eukaryotes serves as a critical regulator of co-transcriptional 18 RNA processing and transcription termination with changes in timing of 19 dephosphorylation by the four CTD phosphatases leading to the production of distinct 20 transcriptional readouts. 21 22 We have previously shown that deletion of the Ser5-P CTD phosphatase Rtr1 1 results in an increase in global Ser5 RNAPII phosphorylation and disruption of 2 termination at specific protein-coding genes [12,48]. In this current work, data shows 3 that RTR1 deletion leads to alterations in CID-containing protein interactions with the 4 RNAPII CTD. Additionally, the interactions between CFIa and CPF are decreased in the 5 absence of Rtr1, suggesting that the timing of CTD dephosphorylation may regulate the 6 formation of stable interactions between the termination machinery, the nascent RNA, 7 and RNAPII. Transcriptome analysis reveals that rtr1∆ cells have decreased levels of a 8 variety of noncoding transcripts. Analysis of Nrd1 occupancy shows that both Nrd1 and 9 RNAPII accumulate at known Nrd1 binding sites in WT cells but that RTR1 deletion 10 causes a decrease in RNAPII and Nrd1 levels at both noncoding and coding genes, 11 suggesting increased elongating RNAPII turnover through termination. This study 12 shows that Rtr1 is globally required for the prevention of premature termination of 13 RNAPII at protein-coding genes by reducing the efficiency of termination through the 14 NNS-dependent termination pathway. Since the NNS termination pathway is known to 15 recruit the RNA exosome to carry out termination-coupled RNA processing and/or 16 decay, the impact of rtr1∆ rrp6∆ was also explored, revealing that the decreases in 17 noncoding RNA levels in rtr1∆ require Rrp6 activity. These findings clearly show that 18 Rtr1 is required to attenuate termination through the NNS pathway. Overall, our findings 19 suggest that precise control of CTD dephosphorylation is required to maintain the 20 balance between elongation and termination at the wide variety of target genes whose 21 transcripts are produced and co-transcriptionally processed by RNAPII. 22

RESULTS 1
Disruption-compensation analysis reveals changes in termination factor 2 interactions in RTR1 deletion cells 3 The phosphorylation status of the RNAPII CTD plays a major role in the 4 regulation of the mechanisms through which transcription termination occurs [49][50][51][52]. 5 We have recently shown that deletion of RTR1 causes global increases in CTD Ser5-P 6 [48] and it has previously been shown that loss of RTR1 results in 3'-end processing 7 defects at the polyA-dependent gene NRD1 [12]. To determine the role of Rtr1 in the 8 regulation of RNAPII interactions with termination factors, we performed Disruption-9 Compensation (DisCo) network analysis. It has been postulated that genetic 10 perturbations can cause edge-specific changes in a protein-protein interaction (PPI) 11 networks. DisCo combines genetic perturbation and in-depth affinity purification-mass 12 spectrometry (AP-MS) studies to obtain unique biological insights into the mechanisms 13 that cause phenotypic changes in gene expression networks. For these studies, we 14 generated dynamic protein-protein interaction networks using Significance Analysis of 15 including a large number of ASTs and other ncRNA transcripts ( Fig. 2A & B). Two-7 hundred and seventy-six transcripts showed upregulation of more than 1.5-fold, many of 8 which were ORF regions. The most significantly reduced transcript was IMD2, a well-9 described target of the NNS termination pathway whose expression is regulated by an 10 intergenic NNS terminator ( Fig. 2A, labeled in green, [34,35,64,65]). The transcript for 11 NRD1 was also significantly reduced in rtr1Δ ( Fig. 2A, labeled in green). NRD1 is known 12 to be regulated by premature RNAPII termination through the Nrd1-Nab3 pathway as a 13 form of auto-regulation [32]. Overall, these data suggest that loss of Rtr1 activity results 14 in the downregulation of a number of different classes of ncRNAs with additional 15 changes in specific mRNAs. These data suggest that the increased Ser5-P RNAPII 16 CTD levels in rtr1Δ cells causes elevated activity of the NNS-dependent termination 17 pathway, stimulating premature transcription termination. 18 19 RNAPII and Nrd1 occupancy was measured genome-wide through chromatin 20 immunoprecipitation followed by exonuclease digestion and genome-wide sequencing 21 (ChIP-exo) as described previously [66]. Considering that Nrd1 does not bind DNA 22 directly, rather it binds nascent RNA and RNAPII CTD repeats at Ser5-P; we predicted 23 that ChIP-exo of Nrd1-TAP would detect regions of DNA bound by RNAPII in complex 1 with Nrd1 (Fig. 3A). To confirm that the binding patterns observed were specific to Nrd1, 2 we compared the Nrd1-TAP ChIP-exo normalized read counts from WT cells to those of 3 Rpb3-FLAG (Fig. 3B). URA8 is known to be regulated by alternative start site selection 4 that is dependent on nucleotide availability and is a known target for NNS-dependent 5 early termination [30,62]. The SOD1 locus is convergent with URA8 and lacks RNA 6 binding sites for Nrd1. Figure 3B illustrates the differences seen in the binding patterns 7 of total RNAPII (Rpb3-FLAG) and Nrd1-bound RNAPII (Nrd1-TAP) when comparing 8 transcripts with high (URA8) and low (SOD1) levels of consensus Nrd1-Nab3 RNA 9 binding sites. The consensus Nrd1 binding site of TTTGTAAAGTT is located 40 nt 10 upstream of the URA8 ATG. The alternative start site is terminated by the NNS pathway 11 in nutrient-rich conditions such as growth in YPD as used in this study. Our ChIP-exo 12 analysis of Rpb3-FLAG shows that RNAPII is localized at the 5'-end of the URA8 gene 13 and throughout the SOD1 coding region (Fig. 3B). The 5'-end localization of RNAPII at 14 URA8 corresponds with the peak of Nrd1 binding in the same area, supporting previous 15 work that found that the majority of URA8 transcript is terminated in early elongation by 16 the NNS pathway, resulting in low-level transcription of full-length URA8 [67]. The levels 17 of Nrd1 association at the SOD1 gene are much lower than at URA8 even though total 18 Rpb3-FLAG occupancy is relatively higher at SOD1 than at URA8 confirming that we 19 are able to obtain selective enrichment of Nrd1 on chromatin using the ChIP-exo 20 approach. 21 22 Upon analysis of the Nrd1 ChIP-exo dataset, a pronounced peak of RNAPII was 1 observed upstream of the location of well-positioned Nrd1-RNAPII complexes (Fig. 3C, 2 arrow 1 vs. arrow 2). These peaks were observed at both protein-coding gene NNS 3 targets such as NRD1 (Fig. 3C) and noncoding genes such as SNR13 (Fig. 3D,  4 compare arrow 1 for upstream peak to arrow 2 for Nrd1-RNAPII peak). This suggests 5 that Nrd1 binding to RNAPII may produce a stably paused RNAPII at strong NNS 6 consensus sites. To further explore these findings, we used previously published Nrd1 7 PAR-CLIP datasets to annotate the top 100 most intense sites and then averaged the 8 Nrd1 and RNAPII intensities surrounding the Nrd1 consensus RNA binding sites [30, 9 31]. In Fig. 3E, we observed a narrow peak of Nrd1-RNAPII complexes located just 10 downstream of the genomic location of the Nrd1 consensus motif (marked with a red 11 line). These data suggest that the pausing of RNAPII occurs just downstream of the 12 genomic location of the Nrd1 consensus sequence, perhaps as the cognate RNA 13 binding motif extends outside the RNA exit channel of RNAPII. A similar peak is 14 observed from the Rpb3 ChIP-exo, although the peak is somewhat 5' shifted, perhaps 15 as an average of the Nrd1 bound and unbound RNAPII populations (Fig. 3F). Globally, 16 Nrd1 was also found to localize to the 5'-end of most mRNA encoding genes and, in 17 agreement with previous studies using ChIP-microarray analysis, the mRNA peak of 18 Nrd1 occupancy occurs 93 +/-3 nucleotides downstream of the annotated mRNA 19 transcription start sites (TSS, Figure S2). 20

RNA Polymerase II and Nrd1 occupancy are reduced at snRNA genes in rtr1Δ 22
Total RNA-Seq analysis revealed changes in a number of ncRNA classes 1 including snRNAs (Fig. 2). This includes snRNA 3'-ends that we have manually 2 annotated as extended transcripts (ETs), which are the regions downstream of snRNAs 3 that are within the zone of termination [1]. Full snRNA transcripts are subsequently 4 subjected to 3'-end processing through the NNS-termination pathway in coordination 5 with the TRAMP complex and the Rrp6-containing RNA exosome [1, 29, 68-72]. 6 Average gene analysis was performed using the ChIP-exo datasets for Rpb3 and Nrd1 7 for the snRNA genes aligned to the TSS with 500 bp of data upstream and 1kb 8 downstream [73]. As shown in Fig. 4A, the average RNAPII signal at snRNA genes is 9 decreased in rtr1Δ cells relative to WT, as is the Nrd1 occupancy. The relative 10 enrichment of Nrd1 to RNAPII is also decreased at snRNA genes, perhaps suggesting 11 that less Nrd1 recruitment is required to mediate termination in cells lacking Rtr1. In 12 fact, the overall reduction in RNAPII levels at snRNAs suggests that termination may 13 occur in a more rapid fashion, thus causing reduced overall RNAPII levels. In support of 14 this hypothesis, the abundance of extended snRNA transcripts is also decreased in 15 rtr1Δ cells in multiple cases, suggesting that termination occurs earlier. If overall snRNA 16 transcription were reduced, we would also expect to measure a decrease in the mature 17 snRNA levels. However, the overall levels of the mature snRNAs for snR62, 32, 189, 18 and 46 are not statistically different than WT, suggesting that the major changes in 19 these RNA occur at their 3'-ends. 20

21
Global levels of RNA Polymerase II and Nrd1 occupancy are altered in RTR1 22

deletion cells 23
Although Nrd1 recruitment is highest at RNAPII target genes containing RNA 1 binding sites for Nrd1-Nab3 (such as URA8), average gene analysis in this study and 2 others has shown that Nrd1 is recruited just downstream of the peak of RNAPII Ser5-P 3 CTD phosphorylation at protein-coding genes [30,37]. At the model protein-coding 4 gene PMA1, RNAPII occupancy is relatively consistent across the entire length of the 5 gene (data not shown). Nrd1 binding, in contrast, peaks in the 5' end of the gene, ~270-6 321 nt past the annotated transcription start site of PMA1. To measure changes that 7 occur in overall RNAPII occupancy in cells +/-Rtr1 we compared the localization of 8 Nrd1-TAP and Rpb3-FLAG at protein-coding genes 1000 nucleotides downstream of 9 the TSS and compared this to histone occupancy data we obtained by MNase-Seq from 10 WT cells (Fig. 5). The overall RNAPII and Nrd1 occupancy observed relative to the 11 annotated transcript end site (TES) +/-200 nucleotides was also determined. Rpb3 12 localization is slightly higher at the 5' end of protein-coding genes in RTR1 deletion cells 13 relative to WT. However, Rpb3 occupancy decreases in rtr1Δ more than in WT cells as 14 RNAPII progresses towards the 3'-end of these genes and at the TES (Fig. 5A). Nrd1 15 levels show a small but consistent decrease in rtr1Δ samples relative to WT across the 16 entire average gene and at the TES. Interestingly, the transition to lower levels of 17 RNAPII occupancy in rtr1Δ relative to WT occurs just following the peak of Nrd1 18 recruitment (overlaid data shown in Figure S3). 19 20 The IMD2 gene is regulated by an intergenic NNS-terminator which maintains 21 basal levels of IMD2 expression until low nucleotide levels stimulate IMD2 expression 22 ( Figure 6A). In rich media conditions, such as those used for these experiments, an 23 upstream IMD2 CUT is produced and terminated by NNS-dependent termination. By 1 ChIP-exo, both RNAPII and Nrd1 can be mapped to the upstream CUT region in WT 2 cells with low levels of RNAPII observed in the IMD2 coding region ( Fig. 6B & C). 3 However, both RNAPII and Nrd1 levels are significantly reduced in rtr1Δ cells similar to 4 observations at snRNA genes ( Fig. 6B & C). In addition to RNA-Seq analysis of WT and 5 rtr1Δ cells we also performed RNA-Seq analysis of rrp6Δ and rtr1Δ rrp6Δ cells to 6 determine the role of the Rrp6-containing exosome in the downregulation of ncRNA 7 transcripts in rtr1Δ cells (Fig. S3A & B). The IMD2 upstream CUT RNA is also 8 decreased in rtr1Δ while it is increased in rrp6Δ (Fig. 6D). The rtr1Δ rrp6Δ cells show 9 similar levels of the IMD2 upstream CUT relative to WT suggesting that rrp6Δ is 10 required to degrade the CUT following NNS-termination in rtr1Δ cells. The IMD2 coding 11 region is also altered by rtr1Δ perhaps due to the close proximity of Nrd1-Nab3 binding 12 sites to the IMD2 TSS (Fig. 6E). IMD2 mRNA levels are ~5-fold decreased in rtr1Δ cells 13 but are not significantly impacted by rrp6Δ with rtr1Δ rrp6Δ cells also showing 14 decreased levels relative to WT cells (Fig. 6E). Northern blot analysis of the IMD2 15 coding region reflected similar ratios to RNA-Seq and confirmed that the terminator 16 over-ride mutant in the Ser5-P CTD phosphatase Ssu72 (ssu72-tov) also showed 17 increased expression of the full length IMD2 transcript (Fig. 6F, [33]). These data 18 suggest that while Ssu72 and Rrp6 are required for proper IMD2 CUT termination and 19 likely degradation, Rtr1 is a negative regulator of NNS termination with knockout of 20 RTR1 leading to decreased IMD2 levels. Since loss of the Rrp6-containing exosome in rtr1Δ leads to recovery of IMD2 1 transcript levels, it is likely that Rtr1 is a positive regulator of noncoding RNA elongation 2 in WT cells through attenuation of NNS-dependent termination. To explore this 3 possibility, a comparison of global RNA-Seq analysis was performed through analysis 4 of rrp6Δ and rtr1Δ rrp6Δ transcriptomes along with WT and rtr1Δ ( Figure S4, Table S4). 5 A focus on the antisense transcripts (ASTs) that were significantly downregulated in 6 rtr1Δ relative to WT (n=104) revealed that the vast majority of the Rtr1-regulated ASTs 7 depend on the Rrp6-containing exosome for their downregulation (Fig. 7). These data 8 further suggest that the downregulation of the Rtr1-regulated ASTs is likely related to 9 increased premature ncRNA termination through the NNS pathway which requires the Northern blot analysis of YKL151C / YKL151C AS expression using single stranded 18 RNA probes shows that YKL151C sense is up-regulated in rtr1Δ as a consequence of 19 YKL151C AS transcript downregulation (Fig. 8B). The opposite effect can be observed 20 in the rrp6Δ, and rtr1Δ rrp6Δ strains showing that decreased NNS-dependent 21 termination leads to YKL151C sense elongation (Fig. 8). In ssu72-tov cells, YKL151C 22 sense transcripts are decreased as well with a coordinate increase in YKL151C 23 antisense although both changes occur to a lower extent than the changes seen in 1 rrp6Δ (Fig. 8B). Furthermore, strand specific RNA-Seq analysis confirmed that 2 YKL151C is upregulated in rtr1Δ while the YKL151C AS is significantly downregulated 3 ( Fig. 8C & D). 4 5 Discussion 6 Our findings are the first to indicate that the Ser5 CTD phosphatase Rtr1 plays a 7 role in premature termination of RNAPII transcripts through increasing the frequency of 8 NNS-dependent termination. Our data indicate that Rtr1 normally limits the co-9 occupancy of Nrd1/RNAPII in WT cells, thereby acting as a key regulator of the NNS-10 dependent termination pathway and the balance between protein-coding and noncoding 11 or cryptic transcription. In addition, these findings suggest that Rtr1 serves as a post- interactions that occur between RNAPII, CFIa, CPF and NNS as a consequence of 20 RTR1 deletion. The decreased interaction between CFIa and RNAPII was evident 21 through both prey-prey correlation analysis and SAINT probability analysis which also 22 revealed that interactions between CFIa and CPF were detected less frequently in rtr1Δ 23 cells. Additionally, interactions between Nrd1 and RNAPII were increased in cells 1 lacking Rtr1 activity supporting our transcriptome-level findings that show that RNA 2 produced from NNS-dependent genes can terminate earlier (Fig. 4) and more efficiently 3 ( Fig. 4-8). Overall, the rtr1Δ transcriptome has a clear decrease in gene expression 4 relative to WT with a bias towards decreases in non-ORF transcripts (Fig. 2). These 5 findings support models suggesting that the CTD is instrumental in the fate of 6 termination choice and adds novel insights suggesting that Rtr1 is a major determinant 7 that regulates alterations in transcription elongation-termination balance. Rtr1 and its 8 human homolog RPAP2 have been implicated in the response to a variety of cellular 9 stresses including changes in carbon source, heat, and ER stress inducers [78-80]. In 10 light of the findings from this study, this suggest that Rtr1-dependent control of 11 premature RNAPII termination could serve as a regulatory point to mitigate cellular 12 stress. 13 14 This work also provides some new insights regarding the mechanisms of Nrd1-15 dependent RNAPII termination. As shown in Figure 3, RNAPII accumulates at well-16 defined Nrd1 binding sites such, in some cases, trailing RNAPII molecules pile up in 17 some cases behind the Nrd1-bound RNAPII (Fig. 3C & D). This suggests that Nrd1 may 18 cause RNAPII pausing as a consequence of RNA binding or perhaps coordinate RNA 19 and RNAPII CTD binding. Furthermore, the affinity purification mass spectrometry 20 studies suggest that Sen1 association with Nrd1-Nab3 occurs at a relatively low 21 frequency. As discussed, SAINT probabilities of proteins that exist within a stable 22 protein complex often exceed a value of 0.95 (Figure 1, [57, 75, 81]). These quantitative 23 findings indicate that Sen1 is not a stable subunit of the Nrd1-Nab3 complex with SAINT 1 probabilities suggesting that it is a transient interaction partner similar to the exosome 2 and TRAMP complexes. A wealth of previous work has established the requirement for 3 Sen1 in addition to Nrd1-Nab3 for RNAPII termination at sn/snoRNA genes in vivo 4 (reviewed in [51]). Taken together, these data strongly suggest that Sen1 is recruited 5 and/or diffuses transiently to terminate Nrd1-Nab3 paused RNAPII rather than 6 functioning as an integral subunit of an NNS complex. In vitro studies have also shown 7 that Sen1 is able to terminate RNAPII in the absence of Nrd1 and Nab3 further 8 suggesting that Sen1 could potentially terminate multiple forms of paused RNAPII. Cells for subsequent biological replicates were grown on different days. 20

Affinity purification of protein complexes 22
Cells were grown to OD600 ≅ 1.5 in YPD broth overnight and collected by 1 centrifugation for 10 minutes at 4000 x g, then washed in H2O and resuspended in 2 25mL TAP lysis buffer per 2.5 grams of pellet (40mM Hepes-KOH, pH 7.5; 10% 3 glycerol; 350mM NaCl; 0.1% Tween-20; fresh yeast protease inhibitors (Sigma; diluted 4 to 1X)). The cells were slowly transferred to liquid nitrogen using a syringe. The frozen 5 cells were pulverized with a mortar and pestle and lysed further in a Waring Blender 6 with dry ice. The frozen lysate was transferred to a new container and allowed to thaw 7 at room temperature. The resulting extract was treated with 100units DNase I and 10µL 8 of 30mg/mL heparin for 10 minutes at room temperature and clarified by centrifugation ChIP-exo and MNase-Seq library construction, EZBead preparation, and Next-18 Gen sequencing were completed using standard methods based on the Life 19 Technologies SOLiD5500xl system as previously described [63]. The resulting 75 nt 20 solid reads were mapped to Saccharomyces cerevisiae sacCer3 reference genome 21 using in-house mapping pipelines that utilizes bfast-0.7.0a [88]. First, rRNA, tRNAs, and 22 poor-quality reads were discarded, and remaining reads were mapped to reference 23 genome sacCer3 and a splice-junction library, respectively. The genomic and splice-1 junction library mapping data were merged at the end. In a second pipeline, the 2 rRNA/tRNA reads were kept and mapped to the reference genome only due to the 3 minimal number of introns in the yeast genome. Read counts per nucleotide were 4 calculated using bamutils from NGSUtils [89]. Differential gene expression was 5 analyzed using edgeR, which has been shown to work well with low replicate numbers 6 [60, 90]. Four biological replicates were used for each genotype in the RNA-Seq 7 analysis. All raw and processed files from the RNA sequencing and ChIP-exo 8 experiments performed for this study have been deposited to Gene Expression 9 Omnibus [GEO] under the accession number GSE87657. 10 11 Genomics data analysis 12 Following data alignment, noncoding transcripts were manually inspected 13 individually using the Integrative Genomics Viewer [91]. To identify ASTs with significant 14 changes in differential expression, the strand was reversed for all sense annotations for 15 the coding region of each ORF-Ts and the text "AS_" was added in front of the ORF-T 16 name. The annotations for the 5' and 3' UTR were not included. These annotations 17 were then used for edgeR analysis and the annotations for ASTs that showed 18 significant changes in rrp6Δ were used for subsequent differential expression analysis 19 to generate the final dataset in Supplementary Table S1. The average gene analysis 20 plots for different RNAPII gene classes were generated using data from two biological Zeta-Probe® blotting membranes by capillary overnight. Transfer efficiency was 8 determined by Methylene Blue staining. Strand specific RNA probes were expressed 9 from a linearized pET-DEST42 (Invitrogen) containing the region of interest in the sense 10 or antisense orientation by T7 transcription (MAXIscript) using 32 P labeled UTP. The 11 radiolabeled probe was purified and then hybridized to the RNA blot at 68˚C overnight. 12 The membranes were then washed with 1xSSC/.1%SDS twice at room temperature 13 and twice with .1xSSC/.1%SDS for 15 minutes at 68°C. Blots were exposed to a 14 phosphorscreen followed by scanning using a phosphorimager (GE Healthcare). 15      Figure S1: STRING network analysis of termination factor complex data using a 6 fold-change cutoff of 5 or more [3]. Networks are included for Pcf11, Nrd1, and 7