Single-molecule characterization of extrinsic transcription termination by Sen1 helicase

Extrinsic transcription termination typically involves remodeling of RNA polymerase by an accessory helicase. In yeast this is accomplished by the Sen1 helicase homologous to human senataxin (SETX). To gain insight into these processes we develop a DNA scaffold construct compatible with magnetic-trapping assays and from which S. cerevisiae RNA polymerase II (Pol II), as well as E. coli RNA polymerase (ecRNAP), can efficiently initiate transcription without transcription factors, elongate, and undergo extrinsic termination. By stalling Pol II TECs on the construct we can monitor Sen1-induced termination in real-time, revealing the formation of an intermediate in which the Pol II transcription bubble appears half-rewound. This intermediate requires ~40 sec to form and lasts ~20 sec prior to final dissociation of the stalled Pol II. The experiments enabled by the scaffold construct permit detailed statistical and kinetic analysis of Pol II interactions with a range of cofactors in a multi-round, high-throughput fashion.

B ecause eukaryotic promoters are intrinsically bidirectional 1 , molecular mechanisms have evolved to repress antisense transcription by promoting termination of the corresponding transcription elongation complex (TEC). The yeast Sen1 helicase, homologous to the human senataxin (SETX) helicase, is responsible for this activity, as well as for the regulation of synthesis and termination of a wide range of noncoding RNAs [2][3][4] . Over the course of the last few years the biochemical 5 and structural 6 properties of this enzyme have come into focus. At the same time, however, the detailed mechanism whereby Sen1 acts to remodel the TEC remains poorly understood, in part due to a lack of quantitative assays allowing one to measure and model the kinetics of TEC remodeling. Here, we develop and utilize a quantitative single-molecule assay reporting on the kinetics of extrinsic eukaryotic transcription termination by the Sen1 helicase.
Single-molecule assays have been used extensively to study transcription by both prokaryotic and eukaryotic RNA polymerases [7][8][9][10][11] . Although formation of prokaryotic TECs from a promoter requires only the classical σ cofactor, formation of eukaryotic TECs from a promoter requires a series of TFs and has typically been characterized by a low efficiency of successful initiation in vitro, raising obstacles to the mechanistic study of eukaryotic transcription at single-molecule resolution 11 . Promoter-dependent Pol II initiation has been sidestepped by pre-assembling Pol II on RNA/DNA scaffolds to form TECs with higher efficiency and study the mechanisms of elongation in high-resolution optical trapping systems 9,12 . However, because these are complex single-round assays in which each DNA molecule can be transcribed only once, relatively low data collection throughput represents a major limitation of such assays. This is particularly true for the study of interactions between extrinsic factors and transcribing Pol II.
Previously, we developed a single-molecule nanomanipulation assay 8,13 , in which DNA molecules were constrained at both ends in a magnetic trap, and using supercoiled DNA substrates we were able to study Escherichia coli RNAP transcription 8,14 and transcription-related processes [15][16][17] with~bp spatial resolution and high data throughput. Here, we present a methodology based on a recyclable scaffold DNA construct 18,19 , which permits multiround analysis of transcription, extending the spatial resolution and data throughput made possible by magnetic trap nanomanipulation to the eukaryotic transcription systems. We use this scaffold construct to characterize structure-function relations in the Pol II elongation complex and study important aspects of elongation, such as the action of the elongation factor TFIIS and the dynamics of R-loop formation. In addition, we employ our system to reveal important features of the transcription termination activity of the conserved helicase Sen1, a key actor in the control of pervasive transcription.

Results
Productive transcription from the scaffold. The scaffold DNA we have developed is a 2.7 kb linear DNA containing a centrally located region of 12 bp in which the two DNA strands are not complementary and a permanent bubble is thus formed (Fig. 1a). Two E. coli his terminators, analogous to the E. coli tR2 terminator previously shown to terminate Pol II transcription 20 , that are positioned flanking the permanent bubble, and oriented in such a way as to capture transcription initiated in either direction from the bubble. This construct thus contains both 144 and 444 bp transcription units, however the sequence of the bubble is such that Pol II can escape only in the direction of the shorter transcription unit when only ATP, UTP, and CTP are present, whereas it can escape bidirectionally when all four NTPs are present ( Supplementary Fig. 1). For the purposes of singlemolecule experimentation the construct is tethered between a magnetic bead and a glass surface, and extended and supercoiled using a magnetic trap. We can thus typically monitor~50 DNAtethered beads in a given field-of-view. Supercoiling the DNA under a low (F = 0.3 pN) extending force allows us to calibrate the topological properties of the scaffold, which are quite similar, if not identical, to those of a regular DNA molecule lacking the permanent bubble ( Supplementary Fig. 2). At this force the torsional response of DNA consists in forming plectonemic supercoils for both positive and negative supercoiling, resulting in a typical decrease in DNA extension of about 50-55 nm/turn. Addition of 1 nM Pol II, 1 μM 2-mer RNA (GpA), 1 mM NTPs, and 1 nM of the TFIIS antibacktracking factor to negatively supercoiled DNA led to discrete changes, or pulses, in DNA extension (Fig. 1b). These pulses reflect the action of individual Pol II molecules and consist in a sudden increase in DNA extension, followed by a constant-extension regime, followed by a sudden return (decrease) of DNA extension to its initial state. Each pulse was characterized by both its extension change, Δl, and its duration, T, and a two-dimensional plot displaying each pulse as a pair of coordinates (Δl, T) is presented Fig. 1c. The extension change is consistent with unwinding of approximately one turn of DNA, as expected for the Pol II transcription bubble, and is discussed in detail below. The distribution of pulse durations could be described by a double-Gaussian distribution with two peaks, the first at 10.0 ± 0.3 s (standard error of mean, SEM) and the second at 26.1 ± 1.6 s (SEM, Fig. 1d). The lifetime distribution was better fit by a double-Gaussian function (reduced chi-square is 0.75, Fig. 1d red line) rather than a single-Gaussian function (reduced chi-square is 4.57, see dashed violet line in Fig. 1d). The two peaks are consistent with Pol II transcribing in either direction from the bubble and terminating when it encounters a his terminator (144 and 444 bp away, respectively). A long tail which is not explained by the double-Gaussian fitting likely corresponds to Pol II molecules, which underwent pausing and backtracking during elongation (see below). The ratio in the two peak times (0.38 ± 0.03) is statistically consistent with the ratio of the lengths of the two transcripts produced from this construct (0.31). Taking into account the two different transcription unit lengths, we infer Pol II transcription rates from the permanent bubble at 13.4 ± 0.5 bp/ s (SEM) and 16.6 ± 1.0 bp/s (SEM) for the short and long transcripts, respectively, consistent with expectations 9,12 . We infer that the rising edge of the pulses indeed correspond to blebbing of a transcription elongation bubble from the permanent bubble as depicted (Fig. 1a), and that the durations of the pulses essentially reflects the length of the transcripts generated by bidirectional transcriptional events initiated from the permanent bubble (see below). We note that the vast majority of pulses which initiate also terminate (N Ter = 407), and a small fraction of readthrough events which initiated without termination showing an increase in DNA extension but never returning to baseline (N nonTer = 62), allowing us to quantify the termination efficiency of Pol II on the strong his terminator from E. coli at 0.87 ± 0.04 (N Ter /(N Ter + N nonTer )). This is comparable to the termination efficiency of E. coli RNAP on his terminator (0.78 ± 0.03, see ref. 21 ). When we replaced the his terminator with the tR2 terminator ( Supplementary Fig. 3F) we obtained a termination efficiency of 0.55 ± 0.05, in agreement with the termination efficiency observed in bulk for Pol II terminating on the tR2 terminator 20 .
Based on this analysis of pulse duration we decomposed the 2D plot into three zones corresponding to synthesis of the short transcript (i.e. events lasting 10 ± 0.7 SD seconds), synthesis of the long transcript (i.e. events lasting 26 ± 0.7SD seconds), and synthesis most likely interrupted by pausing and backtracking (i.e. much longer than 26 s). The distribution of unwinding amplitudes for each zone is shown in the second projection of the 2D plot, Fig. 1e (cyan, blue, and orange curves, respectively). The extent of DNA unwinding in the transcription bubble during synthesis of short and long transcripts is essentially the same (9.9 ± 0.9 bp (SEM) and 9.2 ± 0.3 bp (SEM), respectively), and is significantly larger than that observed for the very long-lived transcription pulses (8.3 ± 0.2 bp (SEM)). These values are within a base-pair of those derived from structural analysis 22 . Overall these results suggest the very long-lived transcription pulses are associated with a deprecated transcription bubble, which may explain why they require so much more time to complete their task.
By also measuring pulse duration using constructs with only 444-base or 806-base transcription units and linearly fitting mean pulse duration vs. transcript unit length, we confirm that pulse lifetime reflects the length of the transcription unit (Fig. 1f) rather than time required for Pol II to completely blebb from the scaffold, or time required for Pol II to terminate transcription. By titrating NTPs from 1 mM to 100 μM, we observed a gradual decrease in the transcription rates of Pol II, which were well fitted with the Michaelis-Menten model giving K m = 230 ± 66 μM and V max = 16.7 ± 2.6 bp/s (Fig. 1f). These results are in quantitative agreement with prior single-molecule measurements of eukaryotic transcription elongation (see refs. 9,12,23 and Table 1).
Furthermore, the experiments discussed so far all included 1 nM TFIIS, which serves to rescue backtracked Pol II 9,24 . In Supplementary Fig. 3 we show that the presence of TFIIS increases the likelihood that Pol II successfully completes transcription in a timely fashion, while its absence increases the fraction of Pol II molecules significantly delayed in their elongation task. Finally, Pol II transcription could also be observed using positively supercoiled DNA, although the initiation frequency was lower than that observed using negatively supercoiled DNA ( Supplementary Fig. 4). We note that a topological change in the bubble associated with Pol II binding could be observed prior to escape from the bubble; independent of NTPs (data not shown), it presumably results from torsional rearrangement of the positively supercoiled bubble upon binding by Pol II. The pulse amplitude was essentially unchanged from that observed on negatively supercoiled DNA, however, the duration of pulses was longer, corresponding to a slower overall rate of bubble escape, transcription, and/or termination for positively supercoiled DNA ( Supplementary  Fig. 4F).
To show that the usefulness of this scaffold construct is not limited to eukaryotic transcription, we analyzed transcription of the Pol2-144-444-his construct but using E. coli core RNA polymerase (core ecRNAP, Supplementary Fig. 5). Transcription pulses were again observed and by applying the 2-D analysis described earlier we obtained an elongation rate on the order of 25 nt/s and a size for the transcription bubble on the order of 9 bases, consistent with prior measurements 8 .
We note that this assay also has use for the study of R-loop formation as demonstrated in Supplementary Fig. 6 and discussed in the Supplementary Text.
Real-time analysis of transcription termination by Sen1. The Sen1 helicase from yeast plays an essential role in the control of pervasive termination by inducing non-coding transcription termination 2,3,5,25,26 . Previous studies have shown that Sen1 can translocate on the nascent RNA and induce dismantling of a stalled elongation complex in an ATP-dependent manner 5,26 . We took advantage of our system to further explore the mechanisms of action of Sen1. We generated stalled elongation complexes by using only ATP, UTP, and CTP, which allow Pol II to initiate and transcribe a 137-base G-less cassette but induce stalling at the first G of a G-stretch. As mentioned earlier the sequence of our DNA templates is designed so that the absence of GTP also restricted bidirectional transcription as this base is made limiting in the scaffold for the initiation of antisense transcription (see Supplementary After washing out free components with buffer containing 1 mM AUC, the stalled Pol II is stable for many hours (Fig. 2a-d). Addition of 500 pM Sen1 and 1 mM ATP caused the DNA extension to return to its baseline value, indicating that Sen1 successfully displaced Pol II from the DNA (Fig. 2b) 3,5,25,26 . Although non-specific interactions between full-length Sen1 and DNA precluded exhaustive experimentation, we found that the helicase domain of Sen1 (Sen1 HD) recapitulated the remodeling activity of Sen1 (Fig. 2c), consistent with previous reports showing that the helicase domain is sufficient for termination in vitro 5,6 . This remodeling activity absolutely required both ATP binding and hydrolysis by Sen1 HD (Fig. 2d).
We next titrate Sen1 HD concentrations against Pol II by incubating 1 nM Pol II, 1 μM 2-mer RNA (GpA), 1 nM TFIIS, 1 mM AUC, 1 mM ATP, and Sen1 HD together. Sen1 HD concentration was progressively increased from 10 to 100 pM and we observed a gradual reduction in the mean lifetime of stalled Pol II (Fig. 3a-c). These data could be fit to a simple The lifetime distribution for this intermediate state (Sen1·sEC)' obeyed single-exponential statistics and did not vary significantly with Sen1 HD concentration (Fig. 3e). Averaging over the four concentrations assayed we obtain a mean of 1/k 3 = 17.5 ± 1.3 s (SD, see legend to Fig. 3e), i.e. k 3 = 0.0572 ± 0.0043 s −1 .
We then used the classical Michaelis-Menten model to analyze the concentration-dependence of the initial time during which stalled Pol II waits before forming this intermediate state (Fig. 3b). Modeling those events which display the intermediate gives K m = 96 ± 45 pM and k′ CAT = 0.0219 ± 0.0097 s −1 , where 1/k′ CAT = 1/k 2 + 1/k s . We thus obtain an estimate for k 2 = 0.0277 ± 0.0156 s −1 (or 1/k 2 = 40 ± 20 s). We note that for k CAT derived from Michaelis-Menten analysis of the total time 1/ k CAT = 1/k S + 1/k 2 + 1/k 3 , and that the value we obtain here for k 2 is consistent with that obtained in the prior analysis of the total time. In other words, the rate-limiting kinetic step obtained by classical Michaelis-Menten analysis of the total duration of events (60 s) is roughly equal to the sum of 1/k 2 , the rate-limiting kinetic step obtained by classical Michaelis-Menten analysis of formation of (Sen1·sEC)' (40 s), and 1/k 3 , the rate at which (Sen1·sEC)' is directly observed to undergo resolution (20 s) (see Table 2 for an overview of fitting parameters).
This suggests that the (Sen1·sEC)' intermediate is always formed, even if it is not directly detected. However, we only observe the intermediate in 1/3 of events (see Table 3). The amplitude and lifetime of the intermediate imply that we are missing another 1/3 of events given the measurement resolution (Gaussian noise on the bead is~20 nm at 31 Hz and the bead cutoff frequency is~4 Hz). It is thus possible that a minority fraction of complexes dissociates directly without formation of an intermediate.
The Prior results obtained from bulk assays have shown that the activity of Sen1 is essentially dependent on the presence of the RNA in the stalled elongation complex 26 . To verify this in the single-molecule assay we repeated our measurements in the presence of Rnase A (0, 50, 100 μg/ml). The fraction of Pol II molecules successfully terminated by Sen1 HD in these conditions decreased from 95% to 30% in the presence of RnaseA ( Supplementary Fig. 8E), in quantitative agreement with ref. 26 .
Finally, to determine whether or not stalled Pol II is fully released from DNA upon remodeling by Sen1, we implemented a translocation assay 16,17 in which the Pol II-DNA system is assembled in the magnetic trap by tethering one end of the DNA to the glass surface and attaching the magnetic bead directly to the Pol II (via a biotin moiety introduced into Pol II) (Fig. 4a). Since a transcribing Pol II molecule can be stalled and backtracked by a~8 pN hindering force 9 , a low extending force (1 pN) was applied to avoid force-induced effects. A Pol II molecule was stalled at +137 in the presence of AUC and restarted transcription towards the glass surface by addition of NTPs (Pol2-G-less-137-T construct, which sustains transcription towards the surface only). Processive Pol II elongation can directly be observed in this assay as the enzyme carries the bead with it towards the surface (Fig. 4b). Different transcription run lengths were observed because some Pol II molecules restart during NTP addition. Pol II stalling can also be accomplished on this substrate (Fig. 4c), and upon addition of either Sen1 FL or Sen1 HD we observe Pol II dissociation as an irreversible and instantaneous loss of the magnetic bead (Fig. 4d).

Discussion
The development of a scaffold transcription construct based on the DNA supercoiling assay provides a way of measuring eukaryotic termination-related processes independently of TFs and in a multi-round fashion allowing for observation of a succession of individual Pol II molecules. This approach is compatible with both eukaryotic and prokaryotic transcription, however, it is not compatible with promoter-dependent studies. Additional work will be required to ascertain whether this approach is compatible with transcription-factor-dependent studies. Here the complex is viewed through the formation and dissolution of the topological transcription bubble universally formed by RNA polymerases. We thus characterize the mechanism and kinetics of dissociation of a stalled elongation complex by the Sen1 helicase responsible for termination of noncoding transcription. Our findings include the observation that Sen1 HD is remarkably affine for the stalled elongation complex and functions very efficiently even at sub-nanomolar concentrations (K M~5 0-100 pM). Nevertheless, the helicase is three times likelier to dissociate spontaneously from stalled Pol II than it is to succeed in displacing Pol II. From a catalytic perspective, Pol II remodeling by Sen1 helicase is slow and involves the formation of a succession of complexes, including an intermediate Sen1-Pol II complex in which the transcription bubble is apparently halfrewound. Sen1 requires ATP binding and hydrolysis to generate this intermediate. The intermediate forms slowly, appearing a full 40 s after binding of Sen1 to Pol II, and has a relatively long lifetime of~20 s before the Pol II dissociates from the DNA. The overall process is thus characterized by slow catalysis on the minute timescale. Taken together these results suggest that the target for Sen1 in vivo is a Pol II molecule which has come to be immobilized. The combination of this method with singlemolecule fluorescence as in the NanoCOSM assay (for nanomanipulation and colocalization of single-molecules 16,17 ) will allow for the direct observation of fluorescently labeled effectors of eukaryotic transcription binding and unbinding from Pol II as it works to transcribe DNA.   Supplementary Table) containing, respectively, an XbaI and an SbfI site separately. This fragment contains a unique, centrally located, and unique KpnI site for insertion, in later cloning steps, of specific transcription units. This 2-kb fragment was subcloned through the XbaI and SbfI sites into the pUC18 polylinker to build the backbone plasmid. The unique EcoRI of pUC18 was destroyed by first digesting with EcoRI, then blunting the overhang with the Quick Blunting Kit (NEB), and finally ligating the blunt ends together with the Quick Ligation Kit (NEB) prior to cloning. We designed and obtained via gene synthesis (Eurofins Genomics) the DNA sequence Pol2-144-444-his (see Supplementary Notes) and which is flanked by KpnI sites (red). This sequence presents a region delimited by unique HindIII (blue) and SpeI (green) sites and into which one can insert the permanently mispaired bubble used as a scaffold for Pol II initiation. This modular region is flanked on either side by the his transcription termination sequence of E. coli (underlined). The sequences to be transcribed and which are contained between the HindIII and his terminator on one side, and the SpeI and his terminator on the other side, are from the E. coli lacZ gene and are 144 and 444 bp long, respectively. The Pol2-144-444-his fragment was cloned into the backbone plasmid via its KpnI sites. The resulting DNA plasmid, named pUC18 Pol2-144-444-his, was purified from an overnight culture of E. coli using the Nucleobond Xtra Midi Plus kit for alkaline lysis and ion exchange (Macherey-Nagel).
We designed and obtained via gene synthesis (Eurofins Genomics) the DNA sequence Pol2-444-444-his (see Supplementary notes). This sequence presents a region delimited by unique HindIII (blue) and SpeI (green) sites, and into which one can insert the permanently mispaired bubble used as a scaffold for Pol2 initiation. It differs from the Pol2-144-444-his sequence described above in that both sequences to be transcribed and which are contained between the HindIII site and his terminator on one side, and the SpeI site and his terminator on the other side, are 444 bp long. Unique AvrII (bold, light green) and MluI (bold, purple) restriction site are also present in this sequence for added modularity. This fragment was cloned into the backbone plasmid via its KpnI sites. The resulting DNA plasmid, named pUC18 Pol2-444-444-his, was purified from an overnight culture of E. coli using the Nucleobond Xtra Midi Plus kit for alkaline lysis and ion exchange (Macherey-Nagel).
We constructed the DNA sequence Pol2-806-806-his via the following procedure. We first PCR amplified two 362 bp fragments from Pol2-444-444-his plasmid using the primers 444_LF and 444_LR (containing, respectively, unique AvrII and SpeI sites shown underlined) and 444_RF and 444_RR (containing, respectively, unique AscI and MluI sites shown underlined).
The first PCR fragment was digested with AvrII and SpeI and cloned into the pUC18 Pol2-444-444-his plasmid via through its AvrII site. The desired orientation of the 362 bp fragment was selected by sequencing and maintained the AvrII site at the 5′ end of the sequence. Then, the second PCR fragment was digested with AscI and MluI and cloned into the pUC18 Pol2-444-444-his plasmid through its MluI site. The desired orientation of the 362 bp fragment was selected by sequencing and maintained the MluI site at the 3′ end of the sequence. The resulting plasmid, named pUC18 Pol2-806-806-his, was expressed and purified as above.
The Pol2-G-less-cassette sequence specified below was designed and obtained by gene synthesis (see Supplementary notes). It contains two KpnI sites (red) and a region flanked by HindIII (blue) and EcoRI (yellow) sites. The sequence also contains unique SpeI (green) and MluI (bold, purple) sites. Downstream of the EcoRI site is a G-less sequence. This sequence was cloned into the backbone plasmid through its KpnI sites. The resulting plasmid, named pUC18 Pol2-G-less cassette, was expressed and purified as above.
We designed and obtained via gene synthesis (Eurofins Genomics) the DNA sequence T5N25-178his (see Supplementary notes). This sequence presents a T5N25 promoter sequence (blue) followed by 178 bp transcription unit and a his terminator. This fragment was cloned into the backbone plasmid via its KpnI sites. The resulting DNA plasmid, named pUC18 T5N25-178his, was purified from an DNA constructs for tethered-DNA supercoiling assays. We prepared three constructs containing a scaffold bubble flanked by a his terminator on each side. The Pol2-144-444-his construct sustains bidirectional transcription from the bubble; in one direction an E. coli his terminator lies 144 bp away; in the other direction a his terminator lies 444 bp away. The Pol2-444-444-his construct differs in that the his terminators are each located 444 bp from the bubble, while for the Pol2-806-806-his construct the his terminators are each located 806 bp from the bubble.
5′ phosphorylated oligonucleotides non-tem1 and tem1 (Eurofins Genomics) were annealed to form a scaffold dsDNA oligo by combining to 50 μM each in 1x PBS, heating to 95 o C for 2 min, and then cooling to room temperature over a 2 h period. The regions underlined in the oligos form an unpaired region after annealing the scaffold, and the ends of the annealed scaffold oligo are compatible with ligation into HindIII and SpeI overhangs.
The Pol2-144-444-his, Pol2-444-444-his, and Pol2-806-806-his constructs for tethered-DNA supercoiling assays were prepared by overnight ligation at room temperature of the scaffold oligo into the pre-digested and purified pUC18 Pol2-144-444-his, Pol2-444-444-his, or Pol2-806-806-his plasmids. The ligation product was then purified with a NucleoSpin Gel and PCR Clean-Up kit (Macherey-Nagel) and then digested with XbaI, SbfI, and AseI (New England Biolabs). The desired DNA fragment (2.7 kb for the Pol2-144-444-his construct, 3 kb for the Pol2-444-444-his construct, and 3.4 kb for Pol2-806-806-his construct) was isolated by gel purification and extraction. The DNA molecules were then ligated to 1 kb DNA fragments modified with multiple biotin groups through the XbaI site and to 1 kb DNA fragments modified with multiple digoxigenin groups through the SbfI site. The modified DNA fragments were synthesized via PCR amplification in the presence of dUTP-biotin and dUTP-digoxigenin, respectively (Strick 2005 nmeths).
The T5N25-178his construct was prepared by digesting the T5N25-178his plasmid with XbaI, SbfI, and AseI (New England Biolabs) and isolating via gel purification and extraction. The produced 2.2 kb DNA were then ligated to 1 kb DNA fragments modified with multiple biotin groups through the XbaI site and to 1 kb DNA fragments modified with multiple digoxigenin groups through the SbfI site.
Stalling constructs for tethered-DNA supercoiling assays. Two constructs containing a scaffold bubble and a G-less transcription cassette were prepared. The Pol2-G-less-137 construct sustains unidirectional transcription from the scaffold bubble and stalls Pol2 137 bp from the bubble. The Pol2-G-less-19 construct sustains unidirectional transcription from the scaffold bubble and stalls Pol2 19 bp from the bubble.
First, plasmid pUC18 Pol2-G-less cassette was digested 3 h with HindIII and EcoRI restriction enzymes (New England Biolabs). The~20 bp fragment contained between the HindIII and EcoRI sites was removed by agarose gel electrophoresis and the longer (~5 kb) fragment extracted from the gel (Macherey-Nagel PCR and Gel Extraction Kit).
5′ phosphorylated oligonucleotides non-tem1 and tem2 (Eurofins Genomics) were annealed to form a second scaffold dsDNA oligo as above. The regions underlined in the oligos form an unpaired region after annealing the scaffold, and the ends of the annealed scaffold oligo are compatible with ligation into HindIII and EcoRI overhangs.
The two stalling constructs (Pol2-G-less-137 and Pol2-G-less-19) for tethered-DNA supercoiling assays were prepared in a similar fashion. For the Pol2-G-less-137 construct, the second scaffold oligo (non-tem1 and tem2) was ligated into the pre-digested Pol2-G-less-cassette plasmid prepared as above. To assemble the Pol2-G-less-19 construct, the first scaffold oligo (non-tem1 and tem1) was ligated into Pol2-G-less-cassette plasmid pre-digested as above but with the HindIII and SpeI restriction enzymes. Remaining steps of the assembly procedure were performed as described above.
Stalling constructs for tethered-Pol II translocation assays. The Pol2-G-less-137-T stalling construct for tethered-Pol II translocation assays was prepared following the same procedures as Pol2-G-less-137 construct preparation, up to and including overnight ligation of the scaffold bubble into the plasmid and purification of the ligation reaction from enzymes using a Macherey-Nagel PCR and Gel Extraction Kit. The purified ligation product was then digested with XbaI and NcoI (the NcoI site is located between the XbaI site and the inserted bubble scaffold) and the target DNA fragments (~4.6 kb) was isolated by gel purification. The target DNA molecules were then ligated to 1 kb DNA fragments modified by multiple digoxigenins via the XbaI site and in the presence of NcoI-restriction enzyme. When the resulting DNA is tethered to an antidigoxigenin-treated glass surface via the digoxigenin groups, transcription from the scaffold directs Pol II towards the glass surface.
Surfaces used for single-molecule experiments were prepared and derivatized with anti-digoxigenin 28 .
Tethered-DNA supercoiling assays. DNA molecules were first attached to 1-μmdiameter streptavidin-coated superparamagnetic beads (Dynabeads MyOne Streptavidin C1; Life Technologies) and then to a glass surface functionalized with anti-digoxigenin. The glass surface was placed atop a homemade magnetic trap which monitors and analyzes the position of the tethered superparamagnetic bead with the PicoJai software package (PicoTwist SARL). Data were collected at video rate (31.0802 Hz) and filtered at 0.5 Hz. The standard deviation for bead fluctuations at 31 Hz was s = 20 nm and the bead cutoff frequency was 4 Hz. Data were processed using custom routines in the Xvin software subsuite of PicoJai. Experiments were performed in standard transcription buffer containing 20 mM K-HEPES pH 7.5, 150 mM K-Glut, 8 mM Mg(Ac) 2 , 0.5 mg/ml BSA, 0.1% w/v Tween 20, 2 mM DTT, and 10 μM ZnCl 2 at 28°C. DNA molecules were extended and torsionally constrained (F = 0.3 pN, where 1 pN = 10 −12 N; superhelical density = −0.027 or −7 turns for negative supercoiling; 0.016 or +4 turns for positive supercoiling).
Continuous-tracking and pulse-chase methodologies were used for tethered-DNA supercoiling assays, showing identical results in terms of quantitative analysis and thus simply representing different levels of optimization of these measurements 8,15 .
The pulse-chase methodology has been described previously 17 . Single-round 'pulse-chase' assays are optimal for measuring the interactions of partner proteins with a single RNA polymerase stalled on DNA, without interference or added measurement noise generated by free RNA polymerase in solution. To stall Pol2 on nanomanipulated, supercoiled DNA, we first added 1 nM Pol II, 1 μM 2-mer RNA, 1 mM AUC, and 1 nM TFIIS and incubated for 2000 s. Longer incubation time was used to stall multiple Pol II molecules on the same DNA construct. We then washed away free components with transcription buffer, and then added 500 pM Sen1 FL or 40 pM Sen1 HD, along with 1 mM ATP, to measure the displacement of stalled Pol2. For measurements assaying the role of ATP binding and hydrolysis in Sen1 activity, 40 pM Sen1 HD was added to stalled Pol2 either in the absence of ATP or in the presence of 1 mM ATPγS. Positive controls of negative controls were conducted on the same molecules by final addition of 40 pM Sen1 HD and 1 mM ATP.
Tethered-Pol II translocation assays. Pol II was stalled on the Pol2-G-less-137-T construct at +137 (the first hybridized base pair downstream of the bubble was counted as +1) by incubating 8 pM DNA with 1 nM biotin-labeled Pol II, 1 μM 2mer RNA, 1 mM AUC containing 1 mM each of ATP, UTP, and CTP, and 1 nM TFIIS at 28°C for 30 min; followed by mixing with streptavidin-coated beads and deposition onto the glass surface. After washing away free components, a 1 pN force was applied to gently extend the DNA molecules. Pol2 transcription was restarted by adding the four nucleotides (1 mM each) alone or with 1 nM TFIIS. For Sen1 termination experiments 500 pM Sen1 HD or Sen1 FL were added along with 1 mM ATP.
Time-traces were analyzed using the PicoJai software suite (PicoTwist SARL) and transcription pulses were manually assigned and analyzed for duration and amplitude.
Bulk assays. DNA oligos (non-tem3 and tem3, Eurofins Genomics, FPLC purified, see Supplementary Table) were mixed in equimolar amounts (10 μM each) in 10 mM Tris pH 7.5, 50 mM NaCl buffer, heated to 95°C for 2 min and cooled slowly to room temperature to form the transcription template containing a permanent unpaired region (or bubble, highlighted in yellow). A final concentration of 0.1 mg/ ml BSA was added after annealing the duplex.