Main

Eukaryotic transcription occurs within chromatin. This poses the long-standing question of how Pol II passes through a nucleosome, the fundamental unit of chromatin1,2. Biochemical3,4,5,6 and single-molecule assays7,8 have shown that Pol II pauses at distinct positions within the proximal half of the nucleosome9,10,11. Pol II enters the nucleosome and stably pauses just before it reaches the nucleosome dyad, at superhelical location (SHL) –1, both in vivo12 and in vitro3,6,13. To efficiently overcome the nucleosome barrier without loss of histones, Pol II requires the help of elongation factors5,6,14,15,16, ATP-dependent chromatin remodeling factors17,18 and histone chaperones15,19,20.

Recently, structures of Pol II−nucleosome complexes5,6,13 have provided a starting point to elucidate the mechanisms of chromatin transcription. These structures showed that transcription into the nucleosome results in partial DNA unwrapping. However, available structures lack the factors that are critical for nucleosome transcription, particularly the chromatin remodeling factor Chd1 (ref. 21) and the histone chaperone FACT22, a heterodimer of subunits Spt16 and Pob3. Chd1 and FACT form a complex23,24 and act in concert25. Binary structures of Chd1 (ref. 26) or FACT27 bound to nucleosomes have revealed how these factors can recognize nucleosome substrates. Here we use a structure−function analysis to investigate how Chd1 and FACT facilitate Pol II passage through a nucleosome and arrive at a mechanistic model of factor-mediated nucleosome transcription.

Results

Chd1 and FACT promote nucleosome transcription

To understand how Chd1 and FACT facilitate nucleosome passage by Pol II, we reconstituted factor-facilitated Pol II transcription through a nucleosome in vitro. We designed an extended nucleosome substrate for the biochemical and structural investigation of nucleosome transcription (Fig. 1a). The substrate consists of a single nucleosome, formed on a modified Widom 601 sequence6, with a 40-base pair (bp) upstream DNA extension (Methods). The DNA extension has a 9-nucleotide (nt) 3′-overhang5 that enables annealing of a fluorescently labeled RNA oligonucleotide and allows for Pol II binding and catalytic RNA extension upon addition of nucleoside triphosphates (NTPs).

Fig. 1: Chd1 and FACT stimulate nucleosome transcription.
figure 1

a, Schematic of nucleosome substrate used for formation of Pol II−nucleosome complexes and RNA extension assays. b, Nucleosome transcription assay shows an increase in full-length product in the presence of FACT, Chd1, or FACT and Chd1. RNA length and corresponding nucleosomal base pairs are indicated. c, Bar graph shows a significant increase in full-length product upon addition of FACT, Chd1, or FACT and Chd1 to Pol II−Spt4/5−TFIIS complexes after 30 min of transcription. n = 3 independent experiments with *P < 0.05, **P < 0.01 with two-tailed t test. d, Mutation of Chd1 to eliminate ATPase activity (D513N) strongly decreases production of full-length RNA product during nucleosomal transcription. n = 3 independent experiments with *P < 0.05, **P < 0.01 with two-tailed t test. Error bars represent ± standard deviation. Unprocessed gel images and derived values for plots are provided as source data and in Supplementary Data 1.

Source data

Full size image

We then used a fluorescence-based RNA extension assay to determine the efficiency of nucleosome transcription in the presence of S. cerevisiae Pol II and various factors (Fig. 1b and Extended Data Fig. 1a). We formed a Pol II elongation complex on the extended nucleosome substrate, provided the elongation factors Spt4/5 (Spt 4 and Spt 5) and TFIIS, and initiated RNA elongation by the addition of 1 mM NTPs. Samples were removed at specific time points, and the RNA products were separated by denaturing gel electrophoresis and quantified. We observed that Pol II paused when its leading edge was located at SHL –5 (Pol II active site at bp 1 of the nucleosome) and again when reaching SHL –1 (Pol II active site at bp 42) (Fig. 1b). These pausing positions correspond to previously described sites where Pol II pauses during nucleosome passage6,13. A small fraction of Pol II could overcome these barriers and transcribe trough the nucleosome, consistent with published observations3,16.

When Chd1 or FACT was added to the reactions containing Pol II, Spt4/5 and TFIIS, we observed a 5- or 7-fold increase in full-length product formation, respectively (Fig. 1b,c). The strong increase in full-length product in the presence of only Chd1 was dependent on the ATPase activity of Chd1, but a weak stimulatory effect was observed also with a catalytically inactive Chd1 variant (Fig. 1d and Extended Data Fig. 1b). Indeed, Chd1 binding to a nucleosome may facilitate transcription because it leads to detachment of two turns of DNA at the Pol II entry site26. The stimulatory effects of Chd1 or FACT depend on the presence of Spt4/5 (Extended Data Fig. 1c,d). When both Chd1 and FACT were included, full-length RNA product formation increased only slightly (Fig. 1b,c), suggesting that the stimulatory effects of Chd1 and FACT are not additive.

Structure of Pol II−Spt4/5−nucleosome−Chd1 complex

To investigate the structural basis for Chd1 and FACT function during Pol II nucleosome passage, we formed a Pol II−Spt4/5−nucleosome complex in the presence of Chd1, FACT and the transition-state analog ADP-BeF3 (Extended Data Fig. 2a–d). Transcription was carried out in the presence of GTP, CTP and UTP, and the complex was purified by size exclusion chromatography followed by mild crosslinking with glutaraldehyde (Methods). We prepared cryo-EM grids, collected a total of 3.76 million particles and obtained a reconstruction at a nominal resolution of 2.9 Å (Table 1, Fig. 2, Extended Data Fig. 3 and Supplementary Video 1). We placed known structures and homology models of Pol II28, Spt4/5 (ref. 29) and the nucleosome core particle30 into the density, adjusted them and modeled the remaining DNA (Extended Data Fig. 4). Additional density was observed for Chd1, but not for FACT, and was fitted with the structure of Chd1 in its post-translocated state26 (Extended Data Figs. 4h,i and 5a). The structure was real-space refined and has good stereochemistry (Table 1).

Fig. 2: Pol II−Spt4/5−nucleosome−Chd1 structure.
figure 2

a, Chd1 domain architecture. Residues at domain boundaries are indicated. Regions modeled in the Pol II−Spt4/5−nucleosome−Chd1 structure are indicated with a black bar. The same color coding is used throughout. b, Two views of the structure related by a 90° rotation. The same color code for Pol II, Spt4/5, histones, metal A, RNA, and template and non-template DNA is used throughout. Spt5N density is shown in surface representation. c, Schematic of the structure indicating key elements.

Full size image
Table 1 Cryo-EM data collection, refinement and validation statistics
Full size table

The structure shows that Pol II adopts the active post-translocated state and has transcribed 27 bp into the nucleosome, as observed biochemically (Extended Data Fig. 2c). The Pol II front edge and active site are located around SHL −3 and −4.5, respectively (Fig. 2b,c and Extended Data Fig. 4a). At this stage, Pol II has unwrapped 45 bp of nucleosomal DNA, exposing the proximal histone H2A−H2B dimer (Extended Data Fig. 4a,g). Spt4/5 binds around the Pol II cleft and the RNA exit channel, as previously observed31,32. The NGN domain of Spt5 is sandwiched between upstream DNA that emerges from Pol II and the nucleosome located downstream. Some density was observed for the negatively charged N-terminal region of Spt5 (Spt5N). Spt5N interacts with the exposed proximal H2A−H2B dimer, likely stabilizing it when DNA is absent (Fig. 2b and Extended Data Fig. 5b).

Chd1 activation and Pol II progression

The structure shows that Chd1 binds the partially unwrapped nucleosome. Chd1 uses its double chromodomain and its ATPase motor domain to contact the nucleosome at SHLs +1 and +2, respectively (Fig. 2 and Extended Data Fig. 4a). In contrast, the DNA-binding region of Chd1 is not observed in our structure and apparently mobile (Fig. 2b). We previously observed that the DNA-binding region binds the second DNA gyre in a nucleosome−Chd1 complex26 (Extended Data Fig. 5c). However, in the structure we present here, the second DNA gyre is no longer available for Chd1 interactions because DNA corresponding to SHL −7 to −5 has been transcribed by Pol II.

These observations explain how Chd1 is activated during transcription elongation (Extended Data Fig. 5c). As Pol II transcribes into the nucleosome, it displaces the DNA-binding region of Chd1 from DNA. The DNA-binding region is known to be an inhibitory domain that restricts Chd1 ATPase activity when engaged with nucleosomal DNA33. Displacement from DNA is predicted to release the inhibitory effect of the DNA-binding region33,34 and thereby activate Chd1. This results in DNA translocation toward the nucleosome dyad and into the Pol II cleft5, thereby facilitating Pol II progression.

Structure of Pol II−Spt4/5−nucleosome−FACT complex

To localize FACT during nucleosome transcription, we reconstituted a transcribing S. cerevisiae Pol II−Spt4/5−nucleosome−FACT complex by withholding Chd1 from the assembly (Extended Data Fig. 2e–h). We employed single-particle cryo-EM to determine the structure of the complex at a nominal resolution of 3.1 Å (Methods), with densities for FACT at lower local resolutions (Extended Data Fig. 6). The high-resolution density observed around the Pol II active site allowed us to unambiguously define the nucleic acid sequence register and revealed that Pol II had stalled with the active site located at bp 17 of the nucleosome (Extended Data Figs. 2g and 7). As in the first structure, Pol II adopts the post-translocated state (Extended Data Fig. 7d).

In this structure, four turns of nucleosomal DNA (SHL −7 to SHL −3) are unwrapped from the histone octamer, and the proximal H2A−H2B dimer is exposed (Fig. 3 and Supplementary Video 2). FACT embraces nucleosomal DNA near the dyad position at SHL +0.5 and contacts the template-strand DNA backbone with the middle domain of subunit Spt16 at SHL −0.5 (Fig. 3b,c). Compared to the structure of the Homo sapiens FACT−nucleosome complex27, the position of FACT appears shifted by one helical turn of DNA toward the side of the nucleosome that is distal to Pol II (Extended Data Fig. 8a). The previously observed position of FACT cannot be adopted in our structure because DNA is present on the distal side of the nucleosome but is absent in the isolated FACT−nucleosome complex structures that were reconstituted with shorter DNA to generate a subnucleosome without Pol II action27. In summary, our structure shows the position of FACT on a nucleosome that was partially unraveled by active Pol II transcription elongation.

Fig. 3: Pol II−Spt4/5−nucleosome−FACT structure.
figure 3

a, Domain architecture of FACT subunits Spt16 and Pob3 (DD, dimerization domain; CTD, C-terminal domain). Residues at domain boundaries are indicated. Regions modelled in the Pol II−Spt4/5−nucleosome−FACT structure are indicated with a black bar. b, Two views of the structure related by a 90° rotation. c, Schematic of the structure indicating key elements.

Full size image

Retention of the proximal H2A−H2B dimer

In our second structure, FACT additionally binds the exposed proximal H2A−H2B dimer using the C-terminal region of Spt16 (Fig. 3 and Extended Data Fig. 7h,i), as observed previously27,35. This finding is consistent with the known role of FACT in the retention of the proximal H2A−H2B dimer during transcription4,36,37. Comparison with our first structure shows that the C-terminal region of Spt16 has replaced Spt5N for H2A−H2B binding. This suggests that Spt5N retains the exposed proximal H2A−H2B dimer until FACT is recruited. In our structure, FACT only binds the partially unwrapped nucleosome, and does not interact with Pol II or Spt4/5, consistent with recent results that showed recruitment of FACT to transcribed nucleosomes, rather than the transcribing complex, in vivo25,38.

Superposition of our two structures results in a clash between Chd1 and the FACT subunit Pob3 (Extended Data Fig. 8b). This indicates that binding of Chd1 and FACT to the nucleosome is mutually exclusive, at least in the highly defined states we trapped in our structures. Nevertheless, Chd1 binds FACT with its flexible N-terminal region (Extended Data Fig. 9), allowing one factor to remain loosely associated with the complex while the other factor binds the nucleosome directly. Additionally, binding of the C-terminal domain of Spt16 to the H2A−H2B dimer is predicted to prevent re-association of DNA with the exposed histone octamer surface, as had been observed in previous cryo-EM studies of Pol II−nucleosome complexes that showed unassigned DNA fragments of unknown origin6. Taken together, these observations support a dynamic mechanism with Chd1 and FACT acting either subsequently or in an alternating manner.

Discussion

Based on our data and previously published data, we propose a model for how Chd1 and FACT mediate nucleosome transcription (Fig. 4 and Extended Data Fig. 10a). When the Pol II−Spt4/5 complex transcribes into a Chd1-bound nucleosome, it would release the DNA-binding region of Chd1. This activates the Chd1 translocase33 and may facilitate Pol II progression. Pol II progression exposes the proximal H2A−H2B dimer, which is temporarily stabilized by Spt5N binding. Further Pol II progression would then generate a binding site for FACT, which can then bind the partially unraveled nucleosome, leading to the displacement of Chd1 and Spt5N. Modeling suggests a ~30-bp window for FACT binding during Pol II progression (Extended Data Fig. 10b). Further Pol II progression would displace FACT from downstream DNA and enable FACT to transfer histones to upstream DNA or to relocate from the proximal to the distal H2A−H2B dimer, thereby preventing loss of histones22,39.

Fig. 4: Model for Pol II passage through a nucleosome.
figure 4

a, Model for Pol II progression through the proximal part of a nucleosomal substrate. For details, please compare text.

Full size image

It remains to be studied whether at each nucleosome both Chd1 and FACT are essential for Pol II transcription. The order of events during nucleosome transcription also remain to be studied further, but it seems that Chd1 functions upstream of FACT because Chd1 can bind complete nucleosomes, whereas FACT can only bind partially unraveled nucleosomes. Chd1 may even recruit FACT because it is known to interact with FACT23,24,40. Indeed, we could confirm the Chd1−FACT interaction biochemically. This finding is consistent with the idea that FACT is recruited near the nucleosome by Chd1 but remains flexible and only binds the nucleosome once DNA is partially unwrapped by transcription. In conclusion, we provide molecular snapshots of the dynamic process of factor-mediated nucleosome transcription and a model for Pol II progression through the nucleosome.

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment.

Molecular cloning

S. cerevisiae Spt4 and Spt5 were cloned into vectors 438-A and 438-C, respectively, using ligation-independent cloning (LIC). Vectors 438-A and 438-C were a gift from S. Gradia (UC Berkeley), Addgene plasmids #55218 and #55220. Using LIC, the two genes were combined on a single 438-series vector. The construct contained Spt5 with an N-terminal 6× His tag followed by a maltose-binding protein tag, and a tobacco etch virus protease cleavage site. Spt4 did not contain tags. Each subunit in the combined vector was preceded by a PolH promoter and followed by a SV40 terminator. S. cerevisiae Spt6 was cloned into vector 438-C using LIC. The construct contained an N-terminal 6× His tag followed by a maltose-binding protein tag and a tobacco etch virus protease cleavage site. A codon-optimized sequence of S. cerevisiae TFIIS for expression in Escherichia coli was cloned into LIC-compatible vector 1-O. Vector 1-O was a gift from S. Gradia (UC Berkeley), Addgene plasmid #29658. The construct contains an N-terminal 6× His tag followed by a Mocr solubility tag and a tobacco etch virus protease cleavage site.

Preparation of protein components

S. cerevisiae Pol II was purified as described previously41. S. cerevisiae Chd1 and FACT were expressed and purified as described26. All insect cell lines used for expression were purchased from Life Technologies (Sf9, Sf21) or from Expression Systems (Hi5) and used as identified by the vendor. Cell lines were not tested for mycoplasma contamination.

S. cerevisiae Spt4 and Spt5 were co-expressed in insect cells using a similar approach as that reported previously42. After harvest, cell pellets were resuspended in lysis buffer 500 (500 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT, 30 mM imidazole, pH 8.0, 0.284 μg ml−1 leupeptin, 1.37 μg ml−1 pepstatin A, 0.17 mg ml−1 PMSF, 0.33 mg ml−1 benzamidine). Cells were lysed by sonication. The cell lysate was subjected to centrifugation (18,000g, 4 °C, 30 min) and ultracentrifugation (235,000g, 4 °C, 60 min). The supernatant containing Spt4/5 was subsequently filtered using 0.2-µm syringe filters (Millipore). The filtered supernatant was applied to a GE HisTrap 5 ml HP (GE Healthcare), pre-equilibrated in lysis buffer. The column was subsequently washed with three column volumes (CV) lysis buffer 500, 3 CV high-salt buffer (1000 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT, 30 mM imidazole, pH 8.0, 0.284 μg ml−1 leupeptin, 1.37 μg ml−1 pepstatin A, 0.17 mg ml−1 PMSF, 0.33 mg ml−1 benzamidine) and 4.5 CV lysis buffer. Bound protein was eluted by gradient over 9 CV to 100% nickel elution buffer (500 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10 % (v/v) glycerol, 1 mM DTT, 500 mM imidazole, pH 8.0, 0.284 μg ml−1 leupeptin, 1.37 μg ml−1 pepstatin A, 0.17 mg ml−1 PMSF, 0.33 mg ml−1 benzamidine) over 9 CV. Fractions containing Spt4/5 were pooled and dialyzed for 16 h against dialysis buffer (300 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT, 30 mM imidazole, pH 8.0, 0.284 μg ml−1 leupeptin, 1.37 μg ml−1 pepstatin A, 0.17 mg ml−1 PMSF, 0.33 mg ml−1 benzamidine). The dialyzed sample was applied to a GE HisTrap 5 ml HP and GE HiTrap Q 5 ml HP column combination. After application of the sample, the HisTrap 5 ml HP was removed, and the HiTrap Q 5 ml HP was washed with 5 CV dialysis buffer. Spt4/5 was eluted using a gradient elution to 100% high-salt buffer over 9 CV. Fractions containing Spt4/5 were concentrated using an Amicon Millipore 15 ml 50,000 MWCO centrifugal concentrator and applied to a GE Superdex 200 Increase 10/300 GL size exclusion column, pre-equilibrated in gel filtration buffer (500 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT). Peak fractions were analyzed with SDS−PAGE. Fractions containing Spt4/5 were concentrated using an Amicon Millipore 15 ml 50,000 MWCO centrifugal concentrator to a concentration of ~20 µM, aliquoted, flash frozen and stored at −80 °C. Typical preparations yielded 300 µg of Spt4/5 from 1.2 L of insect cell culture.

S. cerevisiae Spt6 was expressed in insect cells and subsequently purified with a similar protocol used for H. sapiens SPT6 (ref. 43) with a final concentration of ~60 µM. Typical yields are ~10 mg from 1.2 L of insect cell culture.

S. cerevisiae TFIIS was expressed in E. coli BL21 (DE3) RIL cells grown in LB medium. Cells were grown to an optical density at 600 nm of 0.6 at 37 °C. Expression of TFIIS was subsequently induced with 1 mM isopropyl β-d-1-thiogalactopyranoside at 18 °C for 16 h. Cells were harvested by centrifugation and resuspended in TFIIS lysis buffer (300 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT, 30 mM imidazole, pH 8.0, 0.284 μg ml−1 leupeptin, 1.37 μg ml−1 pepstatin A, 0.17 mg ml−1 PMSF, 0.33 mg ml−1 benzamidine). Cells were lysed by sonication. Two rounds of centrifugation (87,000g, 4 °C, 30 min) were used to clear the lysate. The supernatant was applied to a GE HisTrap 5 ml HP. Affinity purification, dialysis and TEV digest were performed similarly as described for Spt4/5. The dialyzed and TEV-digested sample was applied to a GE HisTrap 5 ml HP, pre-equilibrated in TFIIS lysis buffer. The flow-through containing TFIIS was collected and concentrated using an Amicon Millipore 15 ml 10,000 MWCO centrifugal concentrator and applied to a GE S75 16/600 pg size exclusion column, pre-equilibrated in TFIIS size exclusion buffer (300 mM NaCl, 20 mM Na-HEPES, pH 7.4, 10% (v/v) glycerol, 1 mM DTT). Fractions containing TFIIS were concentrated to ~300 µM, aliquoted, flash frozen and stored at −80 °C. Typical yields were 10 mg from 6 L of E. coli expression culture. Protein identity of all protein components was confirmed by MS.

Nucleosome preparation

Xenopus laevis histones were expressed and purified as described previously26,44. Histone H3 was H3K36Cme3 modified45. DNA fragments for nucleosome reconstitution were generated by PCR essentially as described previously27,36. A vector containing a modified Widom 601 sequence was used as a template for PCR. Super-helical locations were assigned based on previous publications5,6,26,46, assuming direction of transcription from negative to positive SHLs. Large-scale PCR reactions were performed with two PCR primers (forward primer: 5′-ACG AAG CGT AGC ATC ACT GTC TTG-3′; reverse primer: 5′-ATC AGA ATC CCG GTG CCG AGG CCG C-3′) at a scale of 50 ml. Full-length PCR product is reported in Supplementary Data. PCR products were purified using anion exchange chromatography (GE Resoure Q 6 ml) followed by ethanol precipitation. The DNA product was digested with TspRI in 1X CutSmart Buffer (NEB) overnight at 65 °C to generate the 9-nt single-stranded DNA overhang. The digestion product was again purified with anion exchange chromatography, ethanol precipitated and resuspended in water. Nucleosome core particle reconstitution was then performed using the salt-gradient dialysis method44. The resulting nucleosome was purified using a Model 491 Prep Cell (Bio-Rad) and subsequently concentrated to 10–20 µM using an Amicon Millipore 15 ml 50,000 MWCO centrifugal concentrator. Quantification of the purified nucleosome was achieved by measuring absorbance at 280 nm. Molar extinction coefficients at 280 nm were determined for protein and nucleic acid components and were summed to yield a molar extinction coefficient for the reconstituted extended nucleosome.

RNA extension assays

RNA extension assays were performed on the same nucleosomal template substrate used for the structural studies. A 6-FAM 5′-labelled 11-nt RNA (5′-/56-FAM/ rUrUrA rUrCrA rCrUrG rUrC-3′) was employed to monitor the transcription reaction. TFIIS has been added to all transcription reactions to prevent formation of overextended DNA–RNA hybrids and facilitate nucleosome passage5,6. The position of Pol II pausing was assigned by indicating the position of the Pol II active site on the Widom 601 DNA. This provided an unambiguous assignment at nucleotide resolution. Therefore, our pausing sites at bp 1 and bp 42 correspond to the previously described6,13 pause sites, with the Pol II leading edge at SHL −5 and SHL −1, respectively.

All subsequent concentrations refer to the concentration in the final reaction. The final concentrations of buffer components were 130 mM NaCl, 20 mM Na-HEPES, pH 7.4, 3 mM MgCl2, 4% (v/v) glycerol, 1 mM DTT/TCEP. The final volume for each RNA extension reaction was 40 µl. RNA (80 nM), nucleosomal template (80 nM) and S. cerevisiae Pol II (100 nM) were mixed in equimolar ratios and incubated for 5 min on ice. Spt4/5 (120 nM) and additional factors (500 nM each), 10× compensation buffer and water were added to achieve final assay conditions. The sample was incubated for 3 min at 30 °C. Transcription elongation was started by the addition of ATP, CTP, GTP and UTP (1 mM each) and TFIIS (60 nM). Five microliters of the reactions were quenched after 5 min, 10 min and 30 min in 5 µl 2× stop buffer (6.4 M urea, 50 mM EDTA, pH 8.0, 1× TBE buffer) if time courses were performed. Samples were treated with 4 µg proteinase K for 15 min at 37 °C, denatured at 95 °C for 3 min, and separated by denaturing gel electrophoresis (4 µl of sample applied to an 8 M urea, 1× TBE buffer, 12% acrylamide:bis-acrylamide 19:1 gel, run in 0.5× TBE buffer at 300 V for 30 min). RNA extension products were visualized using the 6-FAM label and a Typhoon 9500 FLA imager at an excitation wavelength of 473 nm and emission wavelength range of >520 nm.

Gels were subjected to linear contrast enhancement. Source data for all quantified RNA extension assays are provided in Source Data Fig. 1. All RNA extension assays were performed independently and at least three times. Full-length RNA extension products were quantified using Fiji 1.0. The products were normalized against the total intensity of the respective reaction lane to control for errors during gel loading. Bar charts show mean values and standard deviation as error bars. The following P values were applied *P < 0.05, **P < 0.01. A two-tailed t test was employed to determine statistical significance.

Reconstitution of transcribing Pol II−nucleosome complexes

Complexes for cryo-EM were formed in a final buffer containing 130 mM NaCl, 20 mM Na-HEPES, pH 7.4, 3 mM MgCl2, 1 mM DTT/TCEP, 4% (v/v) glycerol. RNA (480 pmol, same construct as used for RNA extension assays) and nucleosome (120 pmol) were incubated for 5 min on ice. Pol II (120 pmol), Spt4/5 (180 pmol) and Spt6 (180 pmol) were added and incubated for 5 min on ice. Water and compensation buffer were added to reach final buffer conditions, and the sample was incubated for 5 min. Transcription elongation was started by the addition of 1 mM each of GTP, CTP and UTP and 0.4 mM 3′-dATP in the case of the Pol II−Spt4/5−nucleosome−FACT complex. Instead of 3′-dATP, 1 mM ADP-BeF3 was added to the Pol II−Spt4/5−nucleosome−Chd1−FACT complex. TFIIS (108 pmol) was immediately added after addition of NTP.

After 15 min of incubation at 30 °C, Chd1 (180 pmol) and FACT (180 pmol), preincubated with H2A−H2B dimer (180 pmol), or FACT alone (180 pmol), preincubated with H2A−H2B dimer (180 pmol), were added. The transcription reactions were allowed to proceed for an additional 30 min at 30 °C and quenched with EDTA (10 mM final concentration). The samples were subsequently centrifuged and applied onto a Superose 6 3.2/300 Increase size exclusion column (GE Healthcare), equilibrated in complex buffer (100 mM NaCl, 20 mM Na-HEPES, pH 7.4, 3 mM MgCl2, 1 mM TCEP, 4 % (v/v) glycerol). Fractions eluted from the size exclusion column were analyzed using SDS−PAGE and denaturing-urea PAGE (8 M urea, 1× TBE buffer, 12% acrylamide:bis-acrylamide 19:1 gel). Consistent with previous observations5, TFIIS is lost from the elongation complex during size exclusion chromatography. Fractions containing the complex were individually crosslinked and dialyzed against dialysis buffer (100 mM NaCl, 20 mM Na-HEPES, pH 7.4, 3 mM MgCl2, 1 mM TCEP) for 3 h, as described previously47.

The dialyzed complexes with an approximate concentration of 50−100 nM were applied to R2/2 gold grids, Au 200 mesh (Quantifoil). The grids were glow discharged for 100 s prior to application of 2 µl of sample to each side of the grid. After incubation of the sample for 8 s, the grid was blotted and vitrified by plunging into liquid ethane using a Vitrobot Mark IV (Thermo Fisher). The Vitrobot was operated at 4 °C and 100% humidity. A blot force of 5 and blot time between 3 and 5 s were used for the sample preparation. The grids were clipped and subsequently stored in liquid nitrogen before data acquisition.

Cryo-EM analysis and image processing

Cryo-EM data were collected on a Titan Krios II transmission electron microscope (FEI) operated at 300 keV. A K3 summit direct detector (GATAN) with a GIF Quantum Filter with a slit width of 20 eV was used for the data acquisition. Data acquisition was performed using SerialEM at a nominal magnification of 81,000×, corresponding to a pixel size of 1.05 Å per pixel in nanoprobe EF-TEM mode. For the Pol II−Spt4/5−nucleosome−Chd1 dataset, image stacks of 40 frames were collected in counting mode over 2.2 s. The dose rate was ~18.12 e− per Å2 per s for a total dose of 39.87 e− per Å2. For the Pol II−Spt4/5−nucleosome−FACT dataset, image stacks of 40 frames were collected in counting mode over 2.2 s. The dose rate was ~18.30 e− per Å2 per s for a total dose of 40.25 e− per Å2.

Micrographs were stacked and processed using Warp48. CTF estimation and motion correction was performed using Warp48. Particles were picked using an in-house trained instance of the neural network BoxNet2 as implemented in Warp, yielding 3,755,390 particles for the Pol II−Spt4/5−nucleosome−Chd1 dataset and 5,227,093 particles for the Pol II−Spt4/5−nucleosome−FACT dataset. Particles were extracted with a box size of 400 pixels and normalized. Further image processing was performed with cryoSPARC49 and RELION 3.0.7.

For the Pol II−Spt4/5−nucleosome−Chd1 dataset, particles were separated into two subsets and subsequently 3D classified using cryoSPARC49. Particles were selected for the presence of a nucleosome-like density. Selected particles were imported into RELION50. Two subsequent rounds of 3D classification resulted in 247,604 particles with clear nucleosomal density. A 3D refinement of these particles resulted in an overall model of 2.6 Å. The particles were CTF refined, and Bayesian polishing was conducted. A mask encompassing the nucleosome and additional density at SHL +2 was applied to two rounds of 3D classification to select for particles that contain Chd1. This resulted in 30,876 particles with clear density for Chd1 bound to the partially unwrapped nucleosome. Particles were then subsequently 3D refined resulting in map A (EMD-12666) with a resolution of 2.9 Å (FSC threshold 0.143 criterion). The map showed excellent density for Pol II, but the nucleosome−Chd1 part of the map showed flexibility. Therefore, Pol II with Spt4/5 and the nucleosome−Chd1 parts of the maps were individually refined using a mask for Pol II−Spt4/5 or nucleosome−Chd1, respectively. Signal subtraction was applied. This resulted in two masked refinements (Pol II−Spt4/5, map B, EMD-12667; nucleosome−Chd1 with signal subtraction, map C, EMD-12668) with overall improved density. These two masked refinements were combined using the Frankenmap and Noise2map tool set of Warp, resulting in the final composite map (map D, EMD-12449).

For the Pol II−Spt4/5−nucleosome−FACT dataset, particles were separated into two subsets and subsequently 3D classified using cryoSPARC and RELION. Particles were selected for the presence of nucleosome-like densities. The selection resulted in 603,550 particles with clear nucleosome-like density. The selected particles were 3D refined using an angular sampling rate of 7.5° and subjected to further classifications to select for particles with bound FACT. This ultimately resulted in 48,718 particles. These particles were again subjected to 3D classification. After 3D refinement, CTF refinement and Bayesian polishing, the remaining 47,138 particles resulted in a final refinement (map 1, EMD-12669) with an overall resolution of 3.1 Å (FSC threshold 0.143 criterion). To improve densities for the Pol II−Spt4/5 and nucleosome−FACT parts of the map, the particles were subjected to masked refinements (maps 2, EMD-12670 and map 3, EMD-12671). Similar to the Chd1 dataset, the masked refinements were combined using the Frankenmap and Noise2map tools included in Warp, yielding the final map (composite map 4, EMD-12450). Local resolutions of the composite maps were estimated using the RELION built-in tool for the determination of local resolutions.

Model building and refinement

For the Pol II−Spt4/5−nucleosome−Chd1 structure, structures of S. cerevisiae Pol II (PDB 3PO2), X. laevis nucleosome (PDB 3LZ0), Chd1 with ADP-BeF3 (PDB 5O9G) and Spt4/5 (PDB 2EXU) were rigid-body docked into map D and refined using Coot51. DNA from the elongation complex and nucleosomal DNA were connected using Coot. Density in the active site of Pol II allowed unambiguous assignment of DNA register. Surprisingly, the complex had transcribed over the T-less cassette that should stall further elongation. The ADP (Sigma-Aldrich) used in the formation of ADP-BeF3 is reported to be contaminated with up to 2.76% ATP52, possibly providing the required substrate to transcribe past the end of the T-less cassette. Identification of DNA−RNA register was aided by map B.

For the Pol II−Spt4/5−nucleosome−FACT structure, the refined Pol II part of the Pol II−Spt4/5−nucleosome−Chd1 structure was rigid-body docked into the density. Additionally, X. laevis nucleosome (PDB 3LZ0), H. sapiens FACT (PDB 6UPK), and Spt4/5 (PDB 2EXU) were rigid-body docked into map 4 using Coot. S. cerevisiae FACT structures (Spt16, PDB 4IOY; Pob3, PDB 4PQ0) were then superposed onto the docked H. sapiens FACT structure. The dimerization domain of Spt16 and Pob3 were generated using PHYRE2 (ref. 53) and superposed onto the docked FACT structure without any additional manual manipulation. The CTD of Spt16 was modeled de novo as a polyalanine extension of Spt16 with no sequence assignment. DNA from the Pol II part of the structure and the nucleosomal DNA were connected in Coot. The density in the active site allowed for unambiguous assignment of the DNA register, and the nucleic acid sequence was adjusted accordingly. Identification of DNA−RNA register was aided by map 2.

Both atomic models were real-space refined using PHENIX54, with secondary structure restraints against map D (Pol II−Spt4/5−nucleosome−Chd1 model) and map 4 (Pol II−Spt4/5−nucleosome−FACT model).

Figure generation

Figures for structural models were generated using PyMol (version 2.3.4; https://pymol.org/), UCSF Chimera55 and UCSF ChimeraX56. Gel quantification was performed using Fiji, and graphs were generated using GraphPad Prism.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.