High-resolution transcription maps reveal the widespread impact of roadblock termination in yeast

Abstract

Transcription termination delimits transcription units but also plays important roles in limiting pervasive transcription. We have previously shown that transcription termination occurs when elongating RNA polymerase II (RNAPII) collides with the DNA-bound general transcription factor Reb1. We demonstrate here that many different DNA-binding proteins can induce termination by a similar roadblock (RB) mechanism. We generated high-resolution transcription maps by the direct detection of RNAPII upon nuclear depletion of two essential RB factors or when the canonical termination pathways for coding and non-coding RNAs are defective. We show that RB termination occurs genomewide and functions independently of (and redundantly with) the main transcription termination pathways. We provide evidence that transcriptional readthrough at canonical terminators is a significant source of pervasive transcription, which is controlled to a large extent by RB termination. Finally, we demonstrate the occurrence of RB termination around centromeres and tRNA genes, which we suggest shields these regions from RNAPII to preserve their functional integrity.

Keywords: Rap1; pervasive transcription; roadblock termination; transcription readthrough; transcription termination mechanism.

Figure 1. Analysis of the transcripts produced upon transcription termination induced by Rap1

1. Scheme of the reporter used for selecting terminators from naïve sequences. TET_P: doxycycline‐repressible promoter; GAL1 _P: GAL1 promoter. The random sequence (120 nt, red box) was inserted within HSP104 sequences upstream of GAL1 _P . The transcripts produced in the presence (red) or absence (blue) of termination signals are indicated. The readthrough (RT) transcript terminates at a cryptic terminator within the GAL1 promoter. A logo (http://weblogo.berkeley.edu/logo.cgi) derived from the putative Rap1 binding sites found in the selected terminators is shown. The approximate position of the oligonucleotide probe used for Northern blot analysis is indicated by a black arrow.

2. Northern blot analysis of transcripts produced in the presence of a Rap1‐dependent terminator in wt or rrp6∆ cells as indicated. ∆BS: RNAs derived from a construct containing a precise deletion of the Rap1 binding site (BS). A red arrow indicates the position of the short transcript produced at the Rap1 termination site. RT transcripts are indicated by a black arrow.

3. Analysis of the polyadenylation status of transcripts derived from Rap1‐dependent termination. Total, polyadenylated (A+, oligo dT‐selected) and non‐adenylated (A−, oligo dT‐depleted) fractions are analyzed in the strains indicated. Rap1‐terminated and RT transcripts indicated as in (B). U4snRNA and RPS28A RNAs are used as controls for non‐adenylated and adenylated species, respectively.

4. Northern blot analysis of strains containing reporters bearing a Rap1‐dependent or a Reb1‐dependent terminator (clone X3, Colin et al, 2014) as indicated. Reb1 (Reb1‐AA) or Rap1 (Rap1‐AA) anchor away strains were used to deplete either protein by the addition of rapamycin (lanes 7–12, two biological replicates). Red and black arrows indicate short and readthrough transcripts as in (B).

5. Northern blot analysis of RNAs derived from a reporter containing a Rap1‐dependent terminator in a wild‐type (lanes 1–5) or a thermosensitive rsp5‐1 strain (lanes 6–10) grown at different temperatures as indicated. Note that the short RNA (red arrow) mainly represents nascent RNA associated with the roadblocked polymerase.

Source data are available online for this figure.

Figure EV1. 3′ end mapping of Rap1‐terminated transcripts

1. Top: Schematic drawing of the reporter system used for selecting the RB terminators with the position of the insertion and the sequence of the selected clones containing a Rap1 site (purple). In blue the sequence of a constant linker used for constructing the pool. Flanking HSP104 sequences are indicated in red. Bottom: PAGE‐northern blot analysis of RNAs produced by the different constructs after oligonucleotide‐directed RNaseH cleavage at −130 nt from the start of the insertion. The position from the RNase H cleavage point is indicated above the sequences. The presence of several shorter RNAs might reveal the occurrence of termination at sites of RNAPII piling up. All analyses were done in an trf4∆ strain to detect unstable transcripts. All lanes are derived from the same gel; marker M1 is shown twice for clarity.

2. Snapshots showing the RNAPII CRAC and RNAseq signal at intron‐containing genes, illustrating the co‐transcriptional nature of the CRAC signal.

Source data are available online for this figure.

Figure 2. RNAPII occupancy at sites of Rap1 roadblock detected by CRAC analysis

1. A

   RNAPII CRAC profile at a site of roadblock upstream of HYP2 (only the signals on the strand of the annotated features are shown). A peak of CRAC RNAPII signal is visible upstream the site of Rap1 occupancy (blue arrow, ChIP exo data, Rhee & Pugh, 2011) in a wild‐type strain in the presence of rapamycin (dark green track) or Rap1‐AA in the absence of the drug (light green track). The roadblock peak is markedly diminished when Rap1 is depleted from the nucleus by the addition of rapamycin to Rap1‐AA cells (red track). Transcription termination at the RB site is accompanied by the production of a non‐annotated cryptic transcript (uHYP2, gray arrow) with a predominant 3′ end located 13 nt upstream of the Rap1 site (data from Roy et al, 2016). The maximum value of the RNAPII peak is 26 nt upstream of the sequence of the Rap1 site. The position of multiple polyadenylation (pA) sites for HYP2 as defined by 3′‐T‐Fill analysis (Wilkening et al, 2013) is indicated. Note the occurrence of transcriptional readthrough after the roadblock when Rap1 is depleted (inset).

2. B, C

   Same as in (A), with Rap1 sites located between two genes arranged in tandem. The dotted oval underscores the level of polymerase occupancy between the CPF terminator and the roadblock, which is not affected by depletion of the roadblock factor. The maxima of the RNAPII peak are located 33 nt (PIL1) and 15 nt (ALD5) upstream of the first Rap1 binding site.

Figure 3. Metasite analysis of roadblock termination at Rap1 and Reb1 sites

1. Average RNAPII CRAC profile at genomic regions aligned on Rap1 (left panel) or Reb1 (right panel) occupancy sites, in the presence (blue) or absence (red) of the roadblock factor. In both cases, the latter was depleted from the nucleus by the addition of rapamycin. Overlapping purple arrows represent features transcribed downstream of Rap1 or Reb1 occupancy sites. Note that the two panels have a different y‐axis scale due to the average higher expression of Rap1‐dependent genes; a dotted horizontal line marks the same average occupancy for comparison.

2. Same as in (A), using only Rap1 or Reb1 sites located within 300 nt downstream of genes terminated by the CPF pathway. The RNAPII average profile was determined for the wild‐type strain (blue) or an rna15‐2 (green) strain at the non‐permissive temperature of 37°C for the mutant; data from the same cells at permissive temperature have not been plotted, but are available.

Data information: For all panels, the number of sites used is indicated. Rap1 and Reb1 sites used in these analyses are listed in Dataset EV1.

Figure EV2. Roadblock termination functions as a fail safe mechanism to limit constitutive readthrough

1. Aggregate plot showing the average RNAPII CRAC profile at sites of Reb1 occupancy located within 300 nt downstream of genes terminated by the CPF pathway as in Fig 3B. The plot demonstrates that in striking contrast to alteration of the CPF pathway, affecting NNS termination by nuclear depletion of Nrd1 has no significant effects on the accumulation of RNAPII at the site of roadblock. Sites used in these analyses are listed in Dataset EV1.

2. Snapshots illustrating the presence of significant levels of intergenic RNAPII CRAC signals at three tandem gene loci. The transcription initiation sites (TSS, Malabat et al, 2015) detected in the regions shown are indicated on top of the RNAPII CRAC tracks. In all these cases, intergenic initiation cannot be detected, indicating that the intergenic RNAPII signal derives from the constitutive readthrough at the CPF‐dependent terminator of the upstream gene. Note that the levels of the readthrough signal in these cases are comparable to the levels of transcription of the downstream gene.

3. Aggregate plot showing the average RNAPII CRAC profile at sites of Abf1 occupancy located within 300 nt downstream of genes terminated by the CPF pathway as in Fig 3B. The plots show the average RNAPII occupancy in the wild‐type and in rna15‐1 cells grown for 1.5 hours at the non‐permissive temperature. Sites used in these analyses are listed in Dataset EV1.

Figure EV3. Metasite analyses illustrating the profile of RNAPII CRAC signal at various transcription factor binding sites

Aggregate plots showing the profile of RNAPII CRAC signal around sites of binding for the transcription factors indicated. The large peak after each binding site corresponds to downstream events of transcription initiation as for the plots in Fig 3A. The roadblock peak precedes the position of aligned binding sites. Sites used in these analyses are listed in Dataset EV1.

Figure 4. Aggregate RNAPII profile at genes followed by a Rap1 site aligned on the poly(A) site

1. Genes terminated by the CPF pathway and followed by a Rap1 site were aligned on the major poly(A) site and the average RNAPII CRAC signal was plotted for the wild‐type (blue) or Rap1‐AA (red) strain in the presence of rapamycin to induce the nuclear depletion of Rap1 in Rap1‐AA. The termination region, defined by the region displaying an average decrease of the RNAPII signal, is indicated by a yellow rectangle. Note that the signal in this region is not affected by the depletion of the RB. The roadblock peak (RB), indicated by a blue arrow, is broader and smaller in these plots because genomic regions are not precisely aligned on the RB site.

2. As in (A), the average RNAPII profile in wild‐type (blue) and rna15‐2 (green) cells at the non‐permissive temperature was plotted to highlight a bona fide termination defect. To visually appreciate the occurrence of readthrough, we normalized the read counts so that the average RNAPII CRAC signals in gene bodies are comparable in wt and rna15‐2 cells.

3. Boxplots representing the distribution of the ratios between the density of reads in the 100 nt immediately preceding the major poly(A) site (RT) and the density of reads in each ORF (body) in the indicated strains and conditions. The plots were generated by the standard R boxplot function. The central line represent the 50^th percentile. The top and bottom of the box represent the 75^th and 25^th percentile respectively. The bottom whisker is the lowest value still within 1.5 Interquartile range (IQR); the top whisker is the highest values still within 1.5 IQR. The number of sites used for the analysis corresponds to the rap1 sites for which an experimentally defined polyadenylation site could be found within 300 bp upstream of the site (123). A two‐sided t‐test has been used to assess statistical significance. Sites used in these analyses are listed in Dataset EV1.

Figure 5. Analysis of the impact of RNAPII roadblocks in termination of snoRNAs

1. A–D

   RNAPII CRAC and RNAseq profiles at the indicated genomic loci and conditions (only the signals on the strand of the annotated features are shown). The position of the Reb1 or Rap1 occupancy (Rhee & Pugh, 2011) is indicated (light blue filled arrow). The occupancy profile in the presence or absence of the RB factor or Nrd1 has been overlapped for ease of comparison. The regions of the readthrough after the roadblock (dark blue arrow) or after the NNS terminator (red arrow) have been enlarged in the insets.

2. E

   Model of primary (NNS) and secondary (RB) termination at snoRNAs that contain a downstream RB site. The flow of RNAPII is indicated in blue, and the internal arrow indicates the direction of transcription. The sites of NNS and RB termination are indicated, respectively, by red and blue arrows; a site of cryptic or alternative (e.g., CPF‐dependent) termination downstream of the RB is indicated by a black arrow. A low level of natural readthrough at the primary site is indicated by a low schematic flow of polymerases (blue) between the NNS and RB sites, which feeds the RB peak. Under defective NNS termination, this readthrough flux increases, together with the RB (left scheme, dotted line, light green). When the RB is affected (right scheme), only the readthrough due to unblocked polymerases (dotted line, light green) downstream of the RB site is observed, terminating at downstream sites (black arrows). The transcripts produced in the different conditions are indicated by plain or dotted lines, which roughly represent the stability and steady‐state levels of the different species. The colors represent the kind of termination (NNS, RB, or cryptic) that leads to the production of a given species.

Figure 6. Expression of the Rap1 DNA‐binding domain restores expression of genes containing an upstream roadblock

1. A–C

   RNAseq profiles at two tandem features containing an upstream Rap1 roadblock site (RPS24A and RPL11B, A and B). A feature (RPS0A) that depends on Rap1 for transcription activation but for which no upstream RB can be detected is shown in (C) as a control. Only the signals on the strand of the annotated features are shown. Wild‐type Rap1, the Rap1 DNA‐binding domain (Rap1‐DBD), or an empty plasmid was expressed in the Rap1‐AA strain and the endogenous protein was depleted from the nucleus upon addition of rapamycin for the times indicated. RNAseq tracks are shown for the target genes and for neighboring genes as a control. Adjacent loci have been separated by vertical lines when the two features have very different expression levels and different scales have been used.

Figure 7. RNAPII roadblock occurs at centromeres and RNAPIII genes

1. Aggregate plot of RNAPII occupancy (median reads count, PAR‐CLIP data) around centromeres. Centromeres have been aligned on the beginning of the CDEI (top plot) or the CDEIII sequence (bottom plot) and a virtual centromere has been reconstituted by aligning the two plots based on the average length of the centromere. The 5′–3′ direction is indicated by a black arrow for each plot. The structure of the centromere and the interacting factors are schematically shown on the top.

2. Snapshot showing the distribution of polymerases around CEN14. RNAPII CRAC distribution is shown for both wild‐type and rna15‐2 cells at the permissive and non‐permissive temperature for the mutant. A roadblock peak (red arrow) is observed upstream of CDEI in all conditions. Detection of the roadblock upstream of the CDEIII sequence requires increasing readthrough transcription at the upstream gene (CIT1) with the rna15‐2 mutation (bottom track). Cyan arrows indicate the direction of transcription.

3. Metaprofile analysis of RNAPII distribution (median reads count, PAR‐CLIP data) around tRNA genes. Genomic regions were aligned on the transcription start sites (top) or the transcription termination site (bottom) and the plots combined as for centromeres. A scheme of tRNA genes and associated factors is shown on the top of the plots. The 5′–3′ direction is indicated by a black arrow for each plot. Note that reads in the body of tRNAs have been removed because representing contamination from mature tRNAs (Schaughency et al, 2014).

4. Snapshot of RNAPII distribution around tC(GCA)B. The profiles in wt cells grown at 30°C and 37°C have been shown as duplicates, as only minor differences are observed. Roadblock peaks are indicated by red arrows. The footprints of TFIIIB and TFIIIC (Nagarajavel et al, 2013) are shown for comparison. The direction of transcription is indicated by cyan arrows. Reads in the body of tRNAs are from contaminating tRNAs and should not be considered as bona fide RNAPII CRAC signals.

5. RNAPII CRAC profile around the gene coding for 5S rRNA (RDN5) in wt cells at 30°C and 37°C as in (D). Roadblock peaks are indicated by red arrows. The footprints of TFIIIB and TFIIIC are shown. A scheme of the gene and the factors bound is shown in the top of the figure. As for tRNAs, the strong signal in the body of the gene should not be considered as a bona fide RNAPII signal, but contaminating 5S RNA.

Figure EV4. Snapshots showing the RNAPII CRAC signal around representative centromeres

1. A–E

   The structure of each centromere is schematically shown on top and reported, scaled, on the tracks. In (A, B), both strands are shown and the direction of transcription is indicated by a green arrow. In (C–E), only one strand is shown as no significant levels of transcription could be detected on the other strand. RNAPII CRAC tracks derived from rna15‐1 cells are shown at the permissive and non‐permissive temperature to illustrate the effect of increased readthrough transcription at convergent genes, which leads to the accumulation of RNAPII signal at the roadblock.

Figure EV5. Metasite analysis of RNA 3′ ends around centromeres and tRNAs

1. A–D

   Aggregate plots showing the distribution of RNA 3′ ends around centromeres and tRNAs in wt and rrp6∆ cells to detect unstable RNAs. The orientation of centromeres and tRNA relative to transcription is shown by schematic drawings. Note the presence of 3′ ends peaks indicating the occurrence of transcription termination at sites of RB.