To experimentally recapitulate the underlying physical process of barrier crossing, that is nucleosomal DNA unwinding, we mimicked the effect of Pol II passage through the nucleosome using mechanical force. To this end, we adapted a previously described single-molecule DNA unzipping assay (Hall et al., 2009; Rudnizky et al., 2016) in which the two strands of the nucleosomal DNA are held in two optical traps, resulting in a Y-shaped configuration (Figure 1A). We engineered the stem ahead of the fork to consist of two consecutive segments of ‘601’ nucleosome positioning sequence (NPS) (Lowary and Widom, 1998), and a short hairpin loop at the end to prevent tether breaking once all double-stranded DNA (dsDNA) is converted into single-stranded DNA (ssDNA) (Figure 1A). During each experiment, we move the two traps apart at a constant speed of 20 nm/s. When the force reaches ~17 pN, the dsDNA at the stem begins to unzip. When the stem segment does not contain a nucleosome, the DNA unzips following a series of closely spaced transitions occurring in a narrow range of forces between 17–20 pN, dictated by the sequence of the template. Once all dsDNA has been fully converted into ssDNA, the force increases sharply at the hairpin end (Figure 1B). The highly reproducible force-extension signatures from the two consecutive NPS regions allow us to align traces from different unzipping experiments by placing the force and dwell-time of the opening junction at each base pair into register (Figure 1B and Figure 1—figure supplement 1). Upon force relaxation (rezipping), the pattern of closely spaced transitions is reproduced in the inverse sense to that observed in the pulling direction (Figure 1B, Video 1).

Figure 1 with 1 supplement see all

Download asset Open asset

Video 1

Download asset

Next, we repeated the experiment with the second NPS preassembled with a human WT nucleosome. The unzipping force-extension signature of the first NPS matches those obtained above, but that of the nucleosome region deviates significantly due to histone-DNA contacts (Figure 1C and Figure 1—figure supplement 1A, Video 2). Relaxation of the tether results in two identical, sequential rezipping signatures characteristic of the naked DNA in ~75% of the cases (Figure 1B and C), indicating that full DNA unzipping led to complete histone removal. However, in 25% of the cases, when we unzip the same molecule for the second time, the force reaches higher values than with bare DNA but lower than those observed the first time, likely reflecting residual histone-DNA interactions from nucleosomal relics. Since we have no knowledge of nucleosome integrity by the second round of unzipping, we only analyzed the first round of unzipping data for each molecule. Because we moved the trap at a constant speed, the dwell-time of the fork at each position reflects the local histone-DNA interaction strength at that force. Indeed, in these constant pulling velocity experiments, the forces applied to histone-DNA contacts lying deeper in the structure depend on the forces reached previously in undoing earlier contacts. This effect may lead to underestimation of the magnitude of later interactions. Accordingly, we also performed force-jump unzipping experiments on the same constructs in which we suddenly increased the force to 28 pN (this force was chosen to minimize the contribution of the dsDNA sequence to the dynamics of the fork) and held it constant while monitoring the unzipping fork dwell-time at each position along the NPS (Figure 2A). In these experiments, the bare DNA construct unzips to the hairpin end instantaneously, while the fork dwells in the WT nucleosome primarily at 25, 31, and 35 bp into the nucleosome (Figure 2B). The force-weighted residence time histograms (see Methods for details) of the unzipping fork along the entire NPS obtained from these two types of experiments are similar and provide a quantitative description of the barrier to nucleosomal DNA unzipping with single bp resolution that we term the nucleosome topography map (Figures 1E and 2B).

Figure 2 with 1 supplement see all

Download asset Open asset

Video 2

Download asset

The topography maps reveal that the unzipping fork encounters substantial resistance at around 17, 26, 31, 41, 52, 61 and 69 bp into the nucleosome, which correspond to regions of proximal dimer and tetramer interaction with the first half of nucleosomal DNA (Figure 1E). Interestingly, the resistance diminishes significantly after the dyad, suggesting that unzipping the first half of the nucleosome destabilizes the histone-DNA interactions of the second half. As previously observed, major histone-DNA interactions first occur ~55 bp from the dyad and exhibit 5 or 10 bp periodicity as the unzipping fork progresses (Hall et al., 2009; Rudnizky et al., 2016). This observation probably reflects the strong histone-DNA contacts along the DNA minor groove every 10 bp observed in the crystal structure of the nucleosome (Luger et al., 1997). Compared to a previous study (Hall et al., 2009), we noticed a shorter residence time near the nucleosome dyad, which we attribute to differences in pulling geometry, buffer conditions, and/or histone source.

Unzipping of tetrasomes (H3/H4 tetramer assembled on NPS) revealed a substantially diminished barrier compared to the octamer, with unzipping fork dwelling events mostly restricted to locations near the dyad, and much lower maximum force reached during the unzipping process (Figure 1E, Figure 1—figure supplement 1B). These data indicate that the H2A/H2B dimer not only interacts locally with the DNA but it also affects the strength of the H3/H4 tetramer-DNA interaction near the dyad to orchestrate the overall nucleosome stability. As loss or exchange of H2A/H2B dimers has been implicated in important biological processes such as DNA replication (Ramachandran and Henikoff, 2015), transcription (Ramachandran et al., 2017; Kireeva et al., 2002), repair (Ransom et al., 2010), and DNA supercoiling (Sheinin et al., 2013); these findings highlight the potential role of non-local histone-DNA interactions in those processes.

Human H2A.Z and uH2B nucleosomes show altered topography maps when compared to their WT counterparts (Figures 1E and 2B). However, the relative magnitude and distribution of the peaks are differently affected by these two modifications. Specifically, uH2B nucleosomes stabilize the dimer region (16 and 25 bp peaks) with minor effects on the tetramer region (Figures 1E and 2B), suggesting that the attachment of ubiquitin to H2B enhances the barrier height locally. The peaks after the dyad are less pronounced and correspond to regions where nucleosomal DNA interacts with the distal dimer. H2A.Z nucleosomes show enhanced peaks at 41 and 52 bp, while exhibiting much lower heights at 25 and 31 bp (Figures 1E and 2B), indicating a global redistribution of the barrier’s strength along its width. Overall, in H2A.Z nucleosomes the dwell-time peaks are more broadly distributed throughout the first half of the barrier than in their WT or uH2B counterparts, while maintaining their 5 to 10 bp periodicity.

To determine whether this redistribution of the barrier strength reflects features of the individual H2A.Z nucleosomes or the superposition of barriers derived from a heterogeneous molecular population resulting from an enhanced lateral mobility of these nucleosomes, as has been previously suggested (Rudnizky et al., 2016), we counted the number of transitions (rips) per unzipping trace in the nucleosome region. Indeed, the H2A.Z-containing nucleosome has on average one more transition per trace than its WT counterpart (Figure 1—figure supplement 1F) but displays a similar standard deviation. This effect is also evident in constant force unzipping experiments on H2A.Z nucleosomes, in which more dwell-time peaks are observed along the NPS (Figure 2B). To check whether H2A.Z nucleosomes are more mobile compared to WT nucleosomes, we repeatedly unzipped-rezipped single WT or H2A.Z nucleosomes up to the proximal dimer region (maximum force reached to ~25–30 pN). If H2A.Z induces lateral mobility of the nucleosome, the position of initial force rise in the nucleosome region would shift between each partial unzipping-rezipping round, and should be quite evident in our finely registered traces. Surprisingly, in contrast to the report by Rudnizky et al. (2016), we observed no lateral mobility with either WT or H2A.Z nucleosomes, as indicated by the highly reversible and overlapping unzipping signatures in the proximal dimer region (Figure 1—figure supplement 1G and H). Together, these results indicate that the effect of the H2A.Z histone variant in our experiments is not to increase the heterogeneity of the nucleosome population but to significantly redistribute the strength of the barrier, effectively broadening it. This conclusion is also supported by the homogeneous migration of H2A.Z nucleosomes in native gels. (Figure 1—figure supplement 1I).

The distal dimer interaction peaks for H2A.Z nucleosomes are visibly diminished relative to those of WT and uH2B nucleosomes (Figure 1E). Interestingly, we observed an increased cooperativity during the assembly of H2A.Z nucleosomes. As the ratio of octamer to DNA is increased during nucleosome reconstitution, we consistently observed significantly less hexasome formation with H2A.Z than with H2A (Figure 2—figure supplement 1A). It is possible that the global decrease of DNA interaction with the distal dimer observed with H2A.Z nucleosomes could also reflect a more cooperative disassembly during unzipping. To pinpoint what regions within H2A.Z are responsible for its assembly cooperativity, we generated a series of sequence swap mutants between H2A and H2A.Z (Figure 2—figure supplement 1B) (Clarkson et al., 1999). Swapping the sequences of the M3 or the M7 region in H2A.Z with the corresponding sequences in H2A promotes the appearance of hexasomes, indicating decreased cooperativity in assembly (Figure 2—figure supplement 1A). Consistently, the topography map of M3_M7 nucleosomes (an M3 and M7 combined swap mutant) showed intermediate topographical features between H2A.Z and WT nucleosome, with the distal dimer interaction peaks (at 109 and 122 bp) becoming more pronounced than those of H2A.Z nucleosomes (Figure 1E), consistent with the idea that cooperativity in disassembly correlates with this distal interaction. Structurally the M3 region corresponds to the ‘loop 1’ that mediates H2A.Z-H2A.Z interactions within the octamer, and the M7 region corresponds to the ‘docking domain’ that mediates H2A.Z interactions with H3-H4 (Zlatanova and Thakar, 2008). These regions play important roles in the stability of the histone octamer. Thus, unique physical properties of the H2A.Z octamer likely account for the broadened barrier distribution we observed during unzipping.

One unique feature from the nucleosome unzipping traces is the presence of fast, reversible unzipping transitions within the proximal dimer region—spanning the first 40 bp of the NPS—that manifest as ‘hopping’ of force and extension in the unzipping experiments (Figure 3—figure supplement 1A). Hopping in this region is nucleosome-specific, as it is not observed during unzipping of bare NPS DNA (Figure 3—figure supplement 1B). The reversible nature of this transition contrasts with the irreversible transitions observed deeper in the nucleosome, and may indicate a rapid, small scale unzipping that is not large enough to disrupt the structure of the octamer. With nucleosome unzipping occurring at approximately 20–30 pN (Figure 1C–D), to observe hopping at our trap separation rate of 20 nm/s (with trap spring constant of approximately 0.3 pN/nm) requires the system to transition between states on timescales of approximately 2 s or less. To better capture these hopping dynamics, we fixed the trap distance such that the unzipping fork remained localized within the proximal dimer region and monitored the fluctuations of the force and extension (passive mode experiment). Within an empirically determined trap distance range, we obtained equilibrium extension hopping traces (Figure 3A), which show transitions on timescales of ~1 s, aligning with the necessary rate of transitions between states. At each fixed trap separation, we determined the number of unzipped bp to obtain a probability distribution for the length of unzipped DNA (Figure 3B). For both bare and nucleosomal DNA, these distributions show consistent peaks, as expected on a heterogeneous energy landscape where the system populates discrete energy wells separated by transition barriers. We note that certain trap separations allow the system to sample multiple wells, giving rise to a multi-modal distribution (Figure 3B, bottom panel), analogous to the hopping observed in the constant pulling rate unzipping curves. Relative to bare DNA, WT nucleosomes display an additional peak in the distribution of unzipped bps, at approximately 28 bp after the start of the second NPS where the most significant contacts between DNA and the H2A-H2B dimer occur (Figure 3C). This peak implies the existence of a barrier to further unzipping that arises from binding of the DNA to histones, and its position is consistent with the dwell time peak observed in the unzipping traces (Figure 1E, peak at 26 bp).

Figure 3 with 1 supplement see all

Download asset Open asset

Assuming that the observed distributions (Figure 3B and C) correspond to equilibrium Boltzmann statistics, we extracted the energy associated with unzipping of each bp in the proximal dimer region of the nucleosome (Figure 3D, details in Methods). The presence of a strong interaction energy peak at 32 bp into the WT nucleosome and the corresponding energy well preceding this new barrier position account for the appearance of the new preferred state in the distribution of unzipped bps (Figure 3C, peak at 28 bp). Furthermore, the low barrier to rezipping from this state (* in Figure 3D) implies that the dynamics between the two states in the proximal dimer region of the nucleosome (at approximately 18 bp and 28 bp) should be quite rapid, in keeping with the hopping behavior observed in the unzipping curves (Figure 3—figure supplement 1A). Subtracting the unzipping energy of bare DNA from that of the nucleosome, we obtained the additional energy associated with each nucleosome type, providing a measure of the interaction energy between the DNA and the octamer throughout the first quarter of the NPS (Figure 3D, inset). WT, H2A.Z, and uH2B nucleosomes all show strong DNA binding to the proximal dimer region, with a large peak in the interaction energy centered at approximately 35 bp.

Having established the topography of the nucleosomal barrier via mechanical force-induced DNA unwinding, we next set out to determine how this topography manifests in the dynamics of Pol II during transcription through the nucleosome. We used a high-resolution dual-trap optical tweezers instrument together with an improved nucleosomal transcription assay (Figure 4). To accurately gauge the positions of Pol II on the template, we placed a ‘molecular ruler’ in front of the nucleosome (Figure 4A). The ‘molecular ruler’ consists of eight tandem repeats of an artificially designed 64 bp DNA that has a single well-defined, sequence-encoded pause site when transcribed by Pol II in bulk (Figure 4—figure supplement 1A, dashed rectangle) and in single-molecule assays (Figure 4B and Figure 4—figure supplement 1B–D). The repeating pausing patterns of Pol II within the ‘molecular ruler’ generated a periodicity of 21.1 ± 0.3 nm (Figure 4—figure supplement 1B), corresponding to the length of 64 bp DNA under experimental force and buffer conditions. This periodicity serves to align all transcription traces (Gabizon et al., 2018; Herbert et al., 2006) and it also enables the accurate conversion of nanometer distances to basepairs of transcribed DNA (Figure 4—figure supplement 1D).

Figure 4 with 1 supplement see all

Download asset Open asset

We used a bubble initiation method to assemble a stalled biotinylated yeast Pol II elongation complex (Hodges et al., 2009; Bintu et al., 2012) that was ligated downstream to a 2 kb spacer DNA and upstream to the ‘molecular ruler’, followed by a single nucleosome (Figure 4A). The ‘601’ NPS was used to ensure both precise nucleosome positioning and accurate assignment of Pol II positions as it crosses the barrier. Pol II transcription was restarted by supplying a saturating concentration of NTPs (0.5 mM). A Pol II stall site consisting of a short inter-strand cross-linked DNA segment was placed downstream of the NPS (Figure 4A). In these assays, we used force-feedback to maintain a constant 10 pN assisting force throughout the transcription trajectory so that the increase of the distance between the beads serves as an accurate measure of how far Pol II has transcribed. We find this tethering geometry to be superior to prior designs (Hodges et al., 2009; Bintu et al., 2012) because it isolates Pol II from the beads surfaces, thus mitigating photo-damage (Landry et al., 2009). A representative real-time trajectory of Pol II transcribing through the ‘molecular ruler’ followed by bare NPS DNA is shown in Videos 3 and 4.

Video 3

Download asset

Video 4

Download asset

We first obtained traces of Pol II transcribing through bare NPS DNA (Figure 5A and B, black trace). Utilizing the pausing patterns obtained with the ‘molecular ruler’ (Figure 4B, inset), we adapted a recently described algorithm (Gabizon et al., 2018) to align the traces such that the positions of Pol II along the entire NPS are under registry (Figure 4B and Figure 4—figure supplement 1D). The results show that Pol II has a median crossing time within the NPS of 11 s (Figure 6—figure supplement 1A), a pause free velocity of 28.9 ± 3.0 nt-s⁻¹ (Figure 6—figure supplement 1H), and displays very few backtracking events (0.17 per trace) (Supplementary file 2).

Figure 5 with 1 supplement see all

Download asset Open asset

Next, we replaced the bare NPS DNA with an assembled Xenopus WT (xWT) nucleosome on the same template (Videos 5 and 6). As expected, Pol II exhibits a dramatic slow-down, with a median crossing time of 129 s (Figure 6—figure supplement 1B), a pause-free velocity of 11.4 ± 4.1 nt-s⁻¹ (Figure 6—figure supplement 1H), and frequent pausing and backtracking within the NPS region (Figure 4C). 59% of Pol II succeeded in crossing the barrier (Figure 6—figure supplement 1F), which is signaled by its reaching the stall site at the end of the template (Figure 4B). Using the ‘molecular ruler’ to precisely locate Pol II on the template, and after subtracting the enzyme’s footprint (16 bp) (Shao and Zeitlinger, 2017), we obtained a median dwell-time histogram of the leading edge of Pol II along the entire NPS with ±3 bp accuracy (depicted as linear and polar plots in Figure 6A). This dwell-time histogram, which we refer to as a transcriptional map of the nucleosome, illustrates at high resolution and accuracy the height and width of the nucleosome barrier to the elongating Pol II. It complements the topographic map described above, translating it into a ‘functional map’. Our measurements represent a nearly 20-fold resolution and accuracy improvement on previous attempts to obtain a transcriptional map of the nucleosomal barrier, since those experiments could only resolve roughly three barrier regions of ~50 bp each, corresponding to entry, central and exit zones (Hodges et al., 2009; Bintu et al., 2012; Fitz et al., 2016).

Figure 6 with 1 supplement see all

Download asset Open asset

Video 5

Download asset

Video 6

Download asset

Several features of Pol II barrier crossing dynamics emerge from the nucleosome transcriptional map (Figure 6A). First, the effect of the barrier begins immediately after the leading edge of Pol II touches the NPS (3 bp peak); the strength of the barrier is largest at 28 bp and 10–20 bp before the dyad, and is negligible after crossing this pseudo-symmetry axis. Second, we identified a region between 28–64 bp into the NPS where Pol II enters long-lived pauses and backtracks frequently (Figure 4C). These pauses are consistent among different molecules and exhibit a ~ 10 bp periodicity (28, 38, 48, 57, and 64 bp into the NPS). Notably, this region coincides with the region of maximum resistance in the single-molecule unzipping assay (Figures 1E and 2B), implying that the transcriptional barrier encountered by Pol II while crossing the nucleosome reflects, to a first approximation, the barrier mapped by the unzipping assay. Third, some molecules were permanently arrested in this region (Figure 5A,B, gray trace), but those that managed to cross it typically succeeded in reaching the stall site shortly thereafter (Figure 5A). Thus, we speculate that this region (28–64 bp) may play an important regulatory role for barrier crossing by the enzyme. The observed asymmetry of the transcriptional map between both sides of the dyad axis (Figure 6A) may reflect a substantial weakening of the histone-DNA interactions in the presence of the bulky resident enzyme halfway across the barrier. However, the transcriptional map asymmetry is similarly observed in the topographic map (Figure 1E), even though the bulkiness of the enzyme does not play a role in those experiments. It is also possible that the barrier asymmetry reflects changes in the nucleosome integrity by the invading polymerase (Kireeva et al., 2002) or the propagating unzipping fork.

Xenopus histones are traditionally the most-widely used in nucleosome studies because they are well behaved in recombinant form. Since we employed recombinant human histones in the unzipping assays, it was of interest to compare the Pol II transcriptional maps of Xenopus nucleosomes with those obtained with their human counterparts utilizing the same 601 NPS. As seen in Figure 6A, the maps are quite similar except that human nucleosomes confer a significantly higher barrier to transcription (see Figure 5A–B, Videos 7 and 8 for representative traces) than those of Xenopus in the proximal dimer region (Figure 6A, the 28 bp peak in orange and red panels).

Video 7

Download asset

Video 8

Download asset

Interestingly, we observed extensive two state transition dynamics while Pol II is paused at certain sites (frequently at 28 and 63 bp) (Figure 4C, Video 6). While Pol II hopping at 28 bp coincides with hopping of the unzipping fork near this region and probably reflects sampling of alternative nucleosomal states ahead of the enzyme, hopping at a much deeper location into the nucleosome (63 bp) may have a more complex origin. We hypothesize that these latter dynamics may be due to local Pol II-histone interactions or re-wrapping of the nucleosome in front of a backtracked enzyme. Indeed, these hopping dynamics occur exclusively after Pol II enters a deeply backtracked state (Figure 5—figure supplement 1).

We also investigated whether Pol II remains functionally competent after the crossing. The pause-free velocity of Pol II after exiting the nucleosome resumed to 70% (28.6 ± 0.8 nt-s⁻¹) of its value before the crossing (41.1 ± 1.0 nt-s⁻¹). This observation seems to indicate that while the enzyme remains functionally competent, its dynamic state has been affected by the encounter (Figure 6—figure supplement 1H). It remains unknown what changes in the enzyme are responsible for this slowing down and if they are reversible. On the other hand, we probed the integrity of the nucleosome by pulling away the two beads after Pol II crossed the barrier. If the nucleosome survived the traversal by Pol II, it would now lie between the two tethering points, that is the upstream DNA handle and the polymerase. Rarely (<5%) these pulling curves displayed the force-extension signature normally associated with the presence of a nucleosome, suggesting that in situ reassembly of the nucleosome following Pol II traversal was inefficient under our assay conditions. A similar low efficiency of nucleosome reassembly was observed from transcription assays in bulk in the absence of factors added in trans such as FACT (facilitates chromatin transcription) (Hsieh et al., 2013).

Next, we investigated the effects of human H2A.Z and uH2B on Pol II transcription dynamics. Our unzipping assay revealed that H2A.Z and uH2B have distinct effects on the nucleosome barriers (Figures 1E and 2B). Under the same buffer and force conditions, Pol II alone was capable of crossing either H2A.Z (Figure 5C and D, Videos 9 and 10) or uH2B (Figure 5E and F, Videos 11 and 12) nucleosomes. While the crossing probabilities of Pol II through H2A.Z and WT nucleosomes are similar (58% and 59%, respectively) (Figure 6—figure supplement 1F), the distributions of pause sites within the NPS are markedly different (Figure 6A) in that H2A.Z is seen to cause a global redistribution of Pol II pause sites along the entire NPS. Such scattered distribution (3, 28, 36, 59, 66, 87, 101, 115, 125 and 138 bp) differs significantly from that of WT nucleosome (3, 28, 57, 64 bp) where most pauses occur before the dyad. This spreading of the barriers for H2A.Z nucleosomes is a further indication that the force applied to the upstream DNA (which is the same in both experiments) is not the dominant factor responsible for the asymmetry of the WT barrier across the dyad, but that the actual histone-DNA interactions are. Furthermore, these differences are unlikely to stem from H2A.Z nucleosomes being mis-positioned on the starting template, because the first pause site at ~3 bp, where the leading edge of Pol II begins to interact with the nucleosome, is observed in every molecular trajectory obtained with H2A.Z nucleosomes. Instead, we attribute them to broadened distributions of DNA-histone interactions as seen in the topography map for H2A.Z nucleosomes (Figures 1E and 2B, also see discussion on energetic profiles of H2A.Z nucleosome in next section). Clearly, H2A.Z strongly modulates the width of the nucleosomal barrier to transcription.

Video 9

Download asset

Video 10

Download asset

Video 11

Download asset

Video 12

Download asset

In contrast, Pol II transcription through human uH2B nucleosomes has a slightly higher passage probability (76%) than through hWT nucleosomes (Figure 6—figure supplement 1F), although at the expense of longer crossing times (Figure 6—figure supplement 1E). The pause site distribution also resembles that of the hWT or the xWT nucleosomes (Figure 6A), however, the median dwell-time of Pol II at pause sites near the dyad is more than double that for the WT nucleosome, suggesting that uH2B enhanced the height of the nucleosomal barrier to Pol II (Figure 6A). The overall pause-free velocities of Pol II transcription through H2A.Z (11.4 ± 7.1 nt-s⁻¹) and uH2B nucleosomes (12.9 ± 3.4 nt-s⁻¹) are lower than that through hWT (18.2 ± 7.1 nt-s⁻¹) nucleosomes (Figure 6—figure supplement 1H). Consequently, the median crossing times of Pol II through H2A.Z (262 s) and uH2B (304 s) nucleosomes are longer than that through hWT nucleosomes (230 s) (Figure 6—figure supplement 1C–E). Note, however, that pause-free velocity contributes negligibly to crossing time, as the latter is dominated by long pauses. Including pausing, translocation and backtracking, Pol II takes longer to cross uH2B or H2A.Z than hWT nucleosomes (Figure 6B).

Similar to when traversing hWT nucleosomes, Pol II backtracks frequently during transcription through H2A.Z and uH2B nucleosomes (Figure 5—figure supplement 1A,C and E). The average number of backtracks and backtrack depths are similar, but backtrack durations are longer when Pol II transcribes through H2A.Z nucleosomes than through WT and uH2B counterparts (Supplementary file 2), which may again reflect the broader extent of histone-DNA interactions in H2A.Z nucleosomes. Failure to recover from some backtracks seems to contribute to Pol II arrests, the positions of which were more scattered for Pol II transcribing through uH2B and H2A.Z than through WT nucleosomes (Figure 6—figure supplement 1G). Interestingly, backtracked Pol II frequently enters long-lived pauses, some of which are also accompanied by frequent two-state transition dynamics (hopping behavior) (Figure 5—figure supplement 1D and F). During some of the long-lived pauses associated with crossing of uH2B nucleosomes, we also observed three-state hopping behavior (Figure 5—figure supplement 1F) and large hopping transition events (Figure 5F, green trace) of Pol II. As this behavior is rarely observed in WT and H2A.Z traces (Figure 5), we speculate that these dynamics of Pol II reflect the presence of the bulky ubiquitin attachment. Collectively, these data reveal that H2A.Z mainly enhances the width and uH2B mainly enhances the height of nucleosomal barrier to transcription. Consistent with the previously reported higher stability of uH2B nucleosomes (Batta et al., 2011; Chandrasekharan et al., 2009), they pose an overall higher barrier magnitude––especially in the region near the dyad––than their WT counterparts to the passage of polymerase. It is also worth noting that for WT and H2A.Z, but not for uH2B nucleosomes, the transcriptional map replicates the corresponding topography map, suggesting that there could be uH2B-Pol II specific interactions that are not present in the unzipping assay.

Our findings provide direct evidence that H2A.Z or uH2B by themselves affect the crossing dynamics of Pol II; and despite their differential effects on nucleosome topography, they both represent stronger barriers than WT nucleosomes for Pol II.

We use a simplified mechanical model to calculate the expected polymerase dwell times along nucleosomal DNA given a profile of DNA-octamer interaction energies. We build on a previously published Pol II model that includes a mechanical DNA linkage between Pol II and the nucleosome, in addition to nucleosome-Pol II steric interactions (Figure 7A) (Koslover and Spakowitz, 2012). The model assumes that Pol II is unable to actively separate the DNA from the surface of the octamer. Instead, the enzyme behaves as a ratchet that makes progress by rectifying the unwrapping fluctuations of the nucleosomal DNA. The enzyme can also backtrack and diffuse forward to re-engage the 3’-end of the nascent transcript with its active site. The extended model incorporates varying binding energies for the DNA along the nucleosome.

Figure 7 with 1 supplement see all

Download asset Open asset

Polymerase progress along the nucleosomal DNA is modeled as a series of transcription steps and backtracking excursions (Figure 7B), adapting the model of Dangkulwanich et al. (2013) to include interactions with the nucleosome. The individual polymerase steps are assumed to occur on an energy landscape that encompasses both the elastic energy of deforming the unwrapped DNA linker between Pol II and the nucleosome core particle, and the interaction energy between the wrapped DNA and the core particle (Figure 7C). For a given Pol II position, the amount of unwrapped DNA is assumed to fluctuate rapidly around an energy minimum that balances these two contributions. As Pol II steps forward, the linker shortens and the elastic energy increases, with a longer linker entailing a smaller energy increase and thus a more rapid rate constant for polymerase forward motion. The interaction energy profile between the DNA and the histone octamer ahead of Pol II determines the ensemble of linker lengths and, thus, indirectly controls the average time spent by Pol II at each DNA base pair. This model allows us to calculate the expected dwell time of the polymerase at each position.

Accordingly, we use the DNA-octamer interaction energies extracted from equilibrium unzipping data (Figure 3D) to calculate the expected mean dwell times of polymerase in the proximal nucleosomal region. The dwell time peaks resulting from these equilibrium interaction energies approximately correspond to peaks in the experimental dwell time profiles (Figure 7—figure supplement 1). The DNA-octamer interaction energies cannot account for the first dwell time peak (the peak at ~3 bp in Figure 6A) corresponding to the initial encounter between Pol II and the nucleosome. We hypothesize that this peak is the result of additional interactions between Pol II and the histone proteins rather than arising from the difficulty of peeling DNA from the nucleosome.

Because equilibrium interaction energy data was available only for the initial section of the nucleosome (Figure 3), Pol II dwell times further into the nucleosomal sequence cannot be predicted from the data available. Instead, we solve the inverse problem: given the measured Pol II mean dwell times (Figure 7D), we fit the DNA-nucleosome interaction energies (Figure 7E, details in Methods) required to generate this dwell time profile. The dependence of dwell times on the nucleosome binding energy in this model is non-local—time spent at a particular bp depends on the energy required to unwrap a segment of DNA ahead of the polymerase. Consequently, peaks in the dwell time profile arise from interactions that involve both large energy values and span a substantial length of DNA (extended regions of strong binding). For example, the peaks at 29 bp in the dwell time profiles for hWT, uH2B, and H2A.Z nucleosomes (Figure 7D), correspond to large peaks in the interaction energy at 32 bp, consistent with the interaction energy profiles obtained from equilibrium DNA unzipping data (Figure 3D). A two-peaked region of strong binding at roughly 43 bp in the H2A.Z and uH2B nucleosomes gives rise to corresponding double peaks in the dwell time profiles 41 bp into the nucleosomes, with substantially longer pausing times for H2A.Z nucleosomes in this region. An additional broad region of strong binding is seen just before the dyad axis, 62 bp in hWT and uH2B nucleosomes, resulting in the observed Pol II pausing peaks 59 bp into the nucleosome. Interestingly, according to our mechanical model, the predicted interaction energies necessary to generate the observed dwell-time profiles are similar in magnitude for all three nucleosomal types. However, wider peaks of strong binding give rise to increased pause durations in the uH2B transcriptional profile, while a distribution of many narrower peaks accounts for the increased number of pausing sites in the H2A.Z profile (Figure 7D,E).

All DNA modifying enzymes were purchased from New England Biolabs (NEB). Oligonucleotides were purchased from Integrated DNA Technology (IDT). Nucleotide triphosphates were purchased from Thermo Fisher Scientific, and standard salts and buffer components were purchased from Sigma Aldrich. Cloning and DNA template construction follows standard molecular biology techniques unless otherwise noted. The sequences of all oligos used are listed in Supplementary file 1.

DNA arms of the Y structure are prepared by standard PCR reactions using lambda DNA as the template. The left (with BsaI site) and right arm (with biotin) of the Y was amplified using oligos ZC01-ZC02 and ZC03-ZC04, respectively. The left arm was digested with BsaI, and annealed with the right arm to form the Y. The length of the left (after ligation) and right arm dsDNA are 937 and 911 bp, respectively.

The first (for alignment) and second (for loading) NPS were amplified with ZC05-ZC06 and ZC07-ZC08 respectively from the pGM3z-601 plasmid. The first NPS fragment was digested with BsaI, and the second NPS fragment was digested with BsaI/DraIII. The first NPS was ligated with the Y to form Y-alignment. The second NPS was ligated with the end hairpin to form NPS-hp. The end hairpin was pre-folded by heating oligo ZC09°C to 98°C for 2 min, followed by a slow decrease of temperature to 25°C at 1 °C/min. Y-alignment were purified using agarose gel. NPS-hp was purified using native PAGE followed by electro-elution and anion exchange chromatography with HiTrap-Q column.

To couple oligonucleotides to polystyrene beads, ZC10 was hybridized to ZC11 to generate a double stranded oligo containing a phosphorylated 5’-CGGT overhang. Annealing was performed by heating a 1:1 mixture of the oligos in water (0.25 mM each) to 95°C for 10 min, followed by cooling to room temperature. This resulted in the following oligo duplex:

• 5' NH2-TTAATTCATTGCGTTCTGTACACG 3'

• 3’ TTAAGTAACGCAAGACATGTGCTGGC-phos 5’

1 µm diameter carboxylated polystyrene beads were coupled to the prepared double-stranded duplex as follows: 10 µL of 10% (W/V) beads were washed four times with 200 µL coupling buffer (0.1 M MES, pH = 4.7, 150 mM NaCl, 5% DMSO), and dispersed in 20 µL coupling buffer. All centrifugations took place for 5 min at 4500 g. 10 µL of 20 µM double stranded oligo and 6 µL of 2 M 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) were added, followed by vigorous shaking for 2 hr at room temperature. At this point another 10 µL of 2M EDC were added, followed by overnight shaking at room temperature. The remaining active EDC was then quenched by adding 2.5 µL of 1 M glycine, and the beads were washed 5 times with storage buffer (Tris 20 mM pH = 8, 1 mM EDTA, 0.05% Tween 20, 5 mM NaN₃) with 3 min of sonication between washes. The beads (1 µm oligo beads) were finally dispersed at a concentration of 1% (w/v) and stored at 4°C.

The beads were passivated by diluting 6-fold in TE (20 mM Tris, pH 8.0, 1 mM EDTA) and addition of β-casein to a final concentration of 1 mg/ml. The beads were vortexed for 10 min, washed once with TE, dispersed to a concentration of 0.2% (w/v) in TE and stored at 4°C until the experiment.

Recombinant human histones H2A, H2B, H3.3 and H4 were purchased from the Protein Expression and Purification (PEP) Facility at Colorado State University. H2A.Z, H2A/H2A.Z swap mutants and all Xenopus laevis histones were expressed in E. coli, purified and reconstituted into octamers according to standard protocols (Dyer, 2003). uH2B was prepared by crosslinking ubiquitin (G76C) and H2B (K120C) as previously described (Long et al., 2014).

Purified human histone octamers or tetramers were stored in 10 mM Tris, pH 7.6, 1.6 M NaCl, 1 mM EDTA, 1 mM DTT, 20% glycerol at −80°C. The reconstitution of nucleosome was performed using a salt dilution method as described (Dyer, 2003). Briefly, NPS-hp DNA and histone octamers or tetramers were mixed in different molar ratios ranging from 1:0.8 to 1:1.4 and initial salt concentration of the mixture (10 µL) was brought to 2 M NaCl. These reactions were incubated at 30°C in a PCR machine and the following amounts of dilution buffer (10 mM Tris, pH 7.6, 1 mM EDTA, 1 mM DTT, 0.1 mg/mL BSA) were added every 15 min: 3.3, 6.7 5, 3.6, 4,7, 6.7, 10, 30, 20, 100 µL. The reaction products were analyzed by native polyacrylamide gel electrophoresis (4%, acrylamide:bisacrylamide ratio 37.5:1) with 0.5 × TBE plus 5 mM MgCl₂ on ice. The reaction that gave no aggregates and minimal amounts of free DNA was chosen for further concentration using Amicon Ultra centrifugation filters with Ultracel 100K membrane. Concentrated nucleosomes were supplemented with 0.02 % NP40 and stored at 4°C.

Unzipping oligo beads were prepared by ligating 5’-CGGT 1 µm polystyrene oligo beads with Y-alignment DNA, NPS-hp nucleosome (or NPS-hp DNA) using E. coli DNA ligase. The reaction was carried out at 16°C for 2–3 hr. The ligated beads were diluted with TB50 buffer (20 mM Tris, pH 8.0, 50 mM KCl, 5 mM MgCl₂, 1 mM DTT, 10 mM NaN₃, 0.1 mg/mL BSA) to a final bead density of 0.00006% (w/v) and loaded into the tweezing chamber, which was filled with TB50 buffer. Tweezing chamber was pretreated with 5% Pluronic F-127 and 1 mg/mL BSA followed by washing with TB50 buffer prior to each experiment. The 1.26 µm SA beads were diluted directed with TB50 buffer to the same bead density as that of oligo beads.

Optical tweezers experiments were performed in a custom-made dual-trap optical tweezers instrument modified from the design in Comstock et al (Whitley et al., 2017; Comstock et al., 2011). In this configuration, a 1064 nm laser is passed through an acousto-optic deflector, with the laser alternating in position between the two traps every 5 µs. The position of the beads relative to the traps was measured using back focal plane interferometry (Bustamante et al., 2009). Single tethers were formed in situ inside the chamber by trapping an oligo bead in one trap and an SA bead in the second trap, and bringing in close proximity the two traps to allow the biotin on the right arm of the Y to interact with streptavidin (Streptavidin bead).

The pulling experiments used to obtain the average dwell times in Figures 1E and 2B are nonequilibrium in nature, reflecting an average over many out-of-equilibrium encounters with each nucleosome type, and are not direct indicators of equilibrium interaction strength alone. In particular, dwell times to rupture in a given trace will depend on the forces reached during a prior rupture event. Complementary probes of DNA-histone interaction strength (see below) verify that many of the peaks in Figures 1E and 2B are due to strong binding interactions.

Due to the nonequilibrium nature of the experiments used to obtain the residence times in Figures 1E and 2B, we caution against the use of such ‘pulling’ residence times alone to assess interaction strength with high resolution. Data from these pulling experiments is most useful when assessed in conjunction with data from other experimental techniques, such as the equilibrium hopping data in Figure 3, and the transcriptional residence times of Figures 6 and 7.

Once a single tether was confirmed, the trap distance was reset to a value at which the tether force was ~0.4 pN. Unzipping was initiated by moving the two traps apart at a constant speed of 20 nm/s for a total of 875 nm. Under this condition, this pulling speed corresponds to a loading rate of 1.8–2.5 pN/s. Rezipping was conducted at the end of unzipping by bringing the two traps together to the initial trap position at the same speed. The tether was broken manually by increasing the trap distance and calibration was performed as previously described (Berg-Sørensen and Flyvbjerg, 2004). Data was acquired at 800 Hz.

To unzip the construct at constant force, the tether was initially held at ~10 pN and force feedback was turned on to maintain the tether at a constant force of 28 pN. The position and distance between the two beads ware recorded at 800 Hz until the construct was fully unzipped. The force feedback was turned off and the tether was relaxed to ~0.5 pN. For the purpose of aligning the traces and accurately converting nanometer distance to basepairs unzipped, an unzipped trace using constant trap separation speed (as described above) was further obtained from the same tether.

Unzipping was performed at constant trap separation speed of 50 nm/s up to the proximal dimer region where the force starts to rise above the baseline of bare DNA construct, but does not reach 30 pN. The partial unzipping was followed by rezipping to the initial trap position. Typically, this results in a trap movement of ~620 nm. After repeating the unzipping-rezipping cycle for 5–10 times, a final unzipping that unzips all the way to the hairpin end (trap movement of 875 nm) was performed to disrupt the whole nucleosome. A bare DNA trace was also collected immediately following this final unzipping.

To capture hopping of the unzipping fork near the proximal dimer interaction region of the nucleosome, the trap was manually moved apart at a small distance increment of 7.1 nm. At each discrete trap position (passive mode), the distance between the two beads was recorded at 2.5 kHz for 10–300 s. Initially, only fast hopping events characteristic of dsDNA unwinding were present. Once the unzipping fork arrived at the proximal dimer interaction of the nucleosome, additional slow transition events, due to histones binding with and dissociating from dsDNA or ssDNA, could be seen. Recording was terminated when the force reached ~23 pN. The tether was then relaxed to ~0.5 pN. For the purpose of aligning the traces and accurately converting nanometer distance to basepairs unzipped, an unzipped trace using constant trap separation speed (as described above) was further obtained from the same tether.

The plasmid that contains a single repeat sequence (pGEM3z-1×repeat) was first cloned by modifying a pGEM3z-T7A1 plasmid (Gabizon et al., 2018). The construction of the plasmid with eight tandem repeat sequences (pGEM3z-8×repeat) was carried out by following a published protocol (Wu et al., 2016) using BglI, DraIII and EagI restriction sites. This method allows doubling of the repeat number following each cycle of cloning. To ease isolation and purification of the 8 × repeat DNA for future ligation steps, we removed the internal BsaI site in pGEM3z-8×repeat and introduced two BsaI sites flanking the 8 × repeat region by using an overlap PCR strategy. Briefly, two fragments outside the 8 × repeat region of the plasmid were amplified using oligos ZC12-ZC13 and ZC14-ZC15, respectively, and assembled into one fragment using ZC12-ZC15. The assembled fragment, which is devoid of the internal BsaI site, was digested with SapI/EagI and ligated with SapI/EagI digested 8 × repeat fragment from pGM3z-8×repeat. The resulting plasmid pGM3z-8×repeat-2×BsaI contains the 8 × repeat sequence flanking by two BsaI sites, which are included in oligos ZC12 and ZC15. All plasmids containing repeat sequences were transformed and grew in SURE2 competent cells at 30°C. Large amounts of pGM3z-8×repeat-2×BsaI plasmids were purified from 150 mL of SURE2 cells using Zyppy Plasmid Maxiprep Kit.

The 8 × repeat DNA with proper overhangs were digested from pGM3z-8×repeat-2×BsaI using BsaI-HF and purified using 8% native PAGE.

The crosslinked DNA (XLink) used to stall Pol II at the end of the template was prepared by annealing ZC16 and ZC17. The annealed oligos were diluted to 1 µM in TE with 20% DMSO and 50 µM trioxsalen, irradiated by 340 nm UV light for 15 min. Extra trioxsalen (10 µM more) was added and the oligos were irradiated for another 15 min. This procedure was repeated to ensure complete crosslinking. The crosslinked oligos were bound to 1 mL HiTrap Q column (GE Healthcare), washed with 4 mL TE buffer, then 4 ml TE buffer + 250 mM NaCl, eluted with TE + 1 M NaCl, and desalted using Amicon Ultra centrifugation filters with Ultracel 3K membrane. The sequences of the crosslinked oligos are:

• 5’ phos-GGTGTACAGAACGCAATGAATT 3’

• 3’ GGACCACATGTCTTGCGTTACTTAA 5’

NPS DNA (308 bp) that contains the 147 bp ‘601’ NPS was amplified from a pGMZ-3z/601 plasmid using oligos ZC18 and ZC19. The NPS DNA was digested with BsaI/DraIII, purified using HiTrap Q column, and ligated to the crosslinked oligo. The ligation product (NPS-Xlink) was purified using 8% native PAGE.

The 2 kb upstream spacer DNA and 1.5 kb biotin handle DNA were amplified from lambda DNA using oligos ZC20-ZC21, ZC22-ZC23, respectively. PCR products were digested with BsaI and purified using 1% agarose gel. Both the 2 kb spacer and 1.5 kb biotin handle DNA were ligated to 5’-CGGT 1 µm polystyrene oligo beads overnight at 16°C using T4 DNA ligase (NEB). The ligated beads were first washed with TE + 0.5 M KCl + 20 µg/mL β-casein, then washed twice with TE + 20 µg/mL β-casein and resuspended in TE + 20 µg/mL β-casein to a concentration of 0.02% (w/v) for 1.5 kb biotin handle, and 0.2% (w/v) for 2 kb spacer DNA. The ligated beads were stored at 4°C until experiments.

Biotinylated yeast Pol II holoenzyme was expressed, purified and biotinylated as previously described (Kireeva et al., 2003) and was a generous gift of Prof. Craig Kaplan. The stalled Pol II elongation complex was prepared by a bubble initiation method followed by uridine triphosphate (UTP) starvation (Hodges et al., 2009). The sequences for the template DNA strand (TDS), non-template DNA strand (NDS) and short RNA (RNA9) are:

NDS:5’AGGTCTCAGAAGACGCCCGAACAACAGACACAAACACCACGGCCGGCGAGCCAGACACGACCAATTATCTATGTAACTTGCCATATTCAGGATTAT 3’

RNA9:

5’ GACGCCCGA 3’

TDS: 3’TCCAGAGTCTTCTGCGGGCTTGTTGTCTGTGTTTGTGGTGCCGGCCGCTCGGTCTGTGCTGGTTAATAGATACATTGAACGGTATAAGTCCTAATAGTCA-phos 5’

To assemble the stalled complex, TDS was incubated with RNA9, heated to 45°C and cooled down to 20°C at 1 °C/min to form the TDS/RNA9 hybrid. Pol II was added to the hybrid and incubated at room temperature (RT) for 10 min, followed by NDS addition and incubation at 37°C for 15 min. Transcription was initiated by adding ATP/GTP/CTP to a final concentration of 10 µM each and the reaction was incubated at RT for 10 min. If Pol II succeeded in restarting, it will be stalled at the first A site on TDS (bolded and underlined in the sequence above) due to absence of UTP. The relocation of Pol II to the stall site will also expose a BsaI site (underlined above) shielded initially by Pol II and only those complexes in which Pol II succeeded in restarting can be digested and further ligated to the 2 kb upstream spacer. The stalled complex was digested with BsaI-HF at 37°C for 15 min, aliquoted, and stored at −80°C until usage.

Xenopus WT (xWT) nucleosome was reconstituted by salt-dialysis using NPS-Xlink DNA with Xenopus laevis recombinant histone octamer. Human WT (hWT), uH2B and H2A.Z nucleosomes were reconstituted similarly to those used in the single-molecule unzipping assay, except that NPS-Xlink DNA was used. The efficiency of nucleosome reconstitution was assessed by 4% native PAGE. In case where a significant amount of free DNA was present, the nucleosome was further purified by sucrose gradient ultracentrifugation. The nucleosomes were concentrated, supplemented with 0.02 % NP40 and stored at 4°C.

Transcription was performed in TB50 buffer. NTPs concentration was 0.5 mM each of ATP, CTP, GTP, UTP. The 1.5 kb biotins beads were pre-incubated with 0.5 µM neutravidin for 10 min at room temperature and diluted with TB50. Pol II sample beads were prepared by ligating the 1 µm 2 kb spacer DNA beads, Pol II stalled complex, 8 × repeat DNA and nucleosome loaded on NPS-Xlink (or bare NPS-Xlink DNA) using E. coli DNA ligase (NEB) at 16°C for 2 hr. 0.02% of NP40 was also included in the ligation reaction. The overhangs of the various components were optimized such that the ligation occurs at desired orders. The sample beads were diluted with TB50 + 0.02% NP40. The full sequence of the assembled transcription template was available at the end of the document.

To perform the experiment, we first captured a 1.5 kb biotin bead in one trap followed by a Pol II sample bead in the other trap. The two beads were rubbed against each other until a tether is formed. If the tether has expected length, the pair of beads was moved to the experimental position, which is close to the outlet of the NTPs channel. Force feedback was turned on to maintain a constant force of 10 pN and the NTPs channel was opened to start transcription. Data acquisition was started right after force feedback was turned on and terminated once the polymerase reached the end or arrested for more than 300 s without dynamics. To probe the fate of transcribed nucleosomes, force feedback was turned off and the trap distance was reset to a value that gives less than 1 pN force on the tether. The two beads were pulled away from each other by increasing trap distance at a constant speed of 20 nm/s, until the force reaches above 40 pN. From these pulling curves, we rarely detected rips normally associated with nucleosome unwrapping. Trap distance was further increased to break the tether and the beads were calibrated. All transcription data was recorded at 800 Hz.

The 1 × repeat DNA template was amplified from pGM3z-1×repeat plasmid using oligos ZC24-ZC25, digested with BsaI-HF and purified by agarose gel extraction. To determine the main pause site in the 64 bp repeat sequence, Pol II stalled complex was radioactively labeled with [α-³²P]-ATP during initial pulsing. The stalled complex was loaded on streptavidin-coated magnetic beads. The beads were washed with TB130 (20 mM Tris, pH = 8.0, 130 mM KCl, 5 mM MgCl₂, 10 mM DTT, 20 µg/mL BSA) and ligated to the 1 × repeat DNA template using T4 DNA ligase for 1 hr at RT. Transcription was chased by adding 40 µM NTPs mix (ATP, UTP, CTP, GTP, final concentration of 40 µM each) to the stalled complex beads, and terminated by adding 2 × urea stop buffer (8 M urea, 50 mM EDTA) at 10, 20, 40, 60, 120, 300, 600 and 900 s. In parallel, transcription was chased by adding 40 µM NTPs mix together with 50 µM of each type of 3’-deoxynucleotide RNA chain terminators (3’dATP, 3’dCTP, 3’dGTP, 3’dUTP, TriLink Biotechnologies). The reactions were allowed to proceed at room temperature for 10 min before terminated by adding the 2 × urea stop buffer. Samples were extracted with Phenol: Chloroform: Isoamyl alcohol (1:0.9:0.01), precipitated with ethanol and dissolved in 2 × formamide sample buffer (95% formamide, 5 mM EDTA, pH 8.0, with bromophenol blue and xylene cyanol). RNA was resolved on 12% denaturing PAGE, dried and exposed to a phosphorimager screen. Images were captured on the Typhoon imager (GE Healthcare) and processed by ImageJ.

The sequence below shows the stem of the Y structure that contains two consecutive pieces of NPS DNA (bold and italic). The red sequence is the stem of the end hairpin (four bases of the loop not shown here). The 147 bp core ‘601’ NPS is underlined in each segment. The upstream DNA is bridged to the two arms of the Y. The DNA in between are ligation sites. The full length of ssDNA after complete unzipping will be 872 bases (including extra bases from the loop of the hairpin).

(arms)…TTTTGACTACTGACGCGGACATTCAGGAGATGGACCCTATACGCGGCCGCCCTGGAGAATCCCGGTGCCGAGGCCGCTCAATTGGTCGTAGACAGCTCTAGCACCGCTTAAACGCACGTACGCGCTGTCCCCCGCGTTTTAACCGCCAAGGGGATTACTCCCTAGTCTCCAGGCACGTGTCAGATATATACATCCTGTGCATGTATTGAACAGCGACCTTGCAACGATGGACCCTATACGCGGCCGCCCTGGAGAATCCCGGTGCCGAGGCCGCTCAATTGGTCGTAGACAGCTCTAGCACCGCTTAAACGCACGTACGCGCTGTCCCCCGCGTTTTAACCGCCAAGGGGATTACTCCCTAGTCTCCAGGCACGTGTCAGATATATACATCCTGTGCATGTATTGAACAGCGACCTTGCACCCTCCACTCTAGA

Pol II loading sequence (NDS/TDS), 8 × repeat DNA, core ‘601’ NPS and crosslinked DNA (Xlink) are in blue, bold, italic and red, respectively. The transcription starvation site (T in NDS) is bolded and underlined. The NPS-Xlink DNA used for octamer loading is underlined.

AGGTCTCAGAAGACGCCCGAACAACAGACACAAACACCACGGCCGGCGAGCCAGACACGACCAATTATCTATGTAACTTGCCATATTCAGGATTATCAGTAGCGGAAGAGCGAGCTCGGTACCCGATCCAGATCCCGAACGCCTATCTTAAAGTTTAAACATAAAGACCAGACCTAAAGACCAGACCTAAAGACACTACATAAAGACCAGACCTAAAGACGCCTTGTTGTTAGCCATAAAGTGATAACCTTTAATCATTGTCTTTATTAATACAACTtACTATAAGaAGAGACAACTTAAAGAGACTTAAAAGATTAATTTAAAATTTATCAAAAAGAGTATTGACTTAAAGTCTAACCTATAGGATACTTACAGCCATCGAGAGGGACACGGGGAAACACCACCAGCCTCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGTGGATCCGCCGGCCGCAACGATGGACCCTATACGCGGCCGCCCTGGAGAATCCCGGTGCCGAGGCCGCTCAATTGGTCGTAGACAGCTCTAGCACCGCTTAAACGCACGTACGCGCTGTCCCCCGCGTTTTAACCGCCAAGGGGATTACTCCCTAGTCTCCAGGCACGTGTCAGATATATACATCCTGTGCATGTATTGAACAGCGACCTTGCCGGTGCCAGTCGGATAGTGTTCCGAGCTCCCACTCTAGAGGATCCCCGGGTACCGAGCTCGAATTCGCCCTATAGTGAGTCGTATTACAATTCACTGGCCGTCGCACCCTGGTGTACAGAACGCAATGAATT

The 1 × 64 bp repeat sequence is in bold. CAACGCCTCCCGGGCTCACCATCATCCTGACTAGTCTTTCAGGCGATGTGTGCTGGAAAGATCTTATGTCACCCCGTGGATCCGCCGGCCGTCATCACCATCATCCTGACTAGAGTCCTTGGCGAACCGGTGTTTGACGTCCAGGAATGTCAAATCCGTGGCGTGACCTATTCCGCACCGCTGCG

Raw data will be made available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.8sb6h8n.

MATLAB scripts have been deposited in GitHub (Chen, 2019; copy archived at https://github.com/elifesciences-publications/dataprocessDNAunzipping).

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Guillermo Chacaltana and Robert Sosa for technical assistance, and all members of Bustamante laboratory for critical discussion. This work was supported by Howard Hughes Medical Institute, NIH Grants R01GM032543 and R01GM071552 to CB, R01GM098401 to TY, R01GM097260 to CK, the US Department of Energy Office of Basic Energy Sciences Nanomachine Program under Contract DE-AC02-05CH11231 to CB, a grant from Alfred P Sloan Foundation to EFK (FG-2018–10394), and a Burroughs Welcome Fund Collaborative Research Travel Grant to TY; CB is a Howard Hughes Medical Institute Investigator.

© 2019, Chen et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.