Structure of a transcribing Pol II-DSIF-SPT6-U1 snRNP complex

RNA polymerase II (Pol II) facilitates co-transcriptional splicing by recruiting the U1 small nuclear ribonucleoprotein particle (U1 snRNP) to the nascent transcripts. Here, we report the cryo-electron microscopy structure of a transcribing Pol II-U1 snRNP complex with elongation factors DSIF and SPT6. Furthermore, our biochemical analysis revealed that the phosphorylated Pol II carboxyl-terminal domain and SPT6 interact directly with U1 snRNP proteins, facilitating its recruitment to the elongation complex. This multivalent interaction allows efficient spliceosome assembly and ensures transcription processivity. Key words: Transcription, splicing, co-transcriptional splicing, cryo-EM .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 2

In eukaryotes, RNA polymerase II (Pol II) synthesizes the precursor messenger RNA (pre-mRNA), which undergoes several processing steps, including splicing, before serving as a template for protein synthesis. During splicing, the spliceosome removes introns from the pre-mRNA while it is being transcribed by Pol II1, 2, 3, 4, 5. The coupling between transcription and splicing enhances the efficiency and accuracy of splicing. This is exemplified in metazoan genes which often harbour introns that span several kilobases in length6, raising the intriguing question of how distant intron ends are brought into proximity to ensure precise and efficient splicing. The highly repetitive carboxyl-terminal domain (CTD) of Pol II has been proposed to function as a platform to recruit splicing factors7, 8, 9, 10, 11, yet the Pol II CTD itself is insufficient to stimulate splicing12. A direct interaction between Pol II and U1 snRNP, independent of the Pol II CTD, was revealed by a cryo-electron microscopy (cryo-EM) structure13. U1 snRNP is the first spliceosome component recruited to the pre-mRNA to recognize the 5´ splice site (5´ SS)14. The direct Pol II-U1 snRNP interaction retains the 5´ SS near the RNA exit site of Pol II, thereby bridging the 5´ SS to the branch point sequence and 3´ splice site as they emerge from Pol II. This intron loop model is further supported by CRISPR interference experiments revealing that the nascent 5´ SS base paired with U1 snRNA is tethered to Pol II during intron synthesis15. Nevertheless, it remains unknown how transcription elongation factors affect U1 snRNP recruitment to an elongating Pol II. Release of Pol II into productive elongation requires the positive transcription elongation factor b (P-TEFb)16, 17 which phosphorylates the Pol II CTD, the DRB sensitivity inducing factor (DSIF, a heterodimer of SPT4 and SPT5) and the negative elongation factor (NELF)18, 19, 20, 21. This change in phosphorylation state

in dissociation of NELF and recruitment of elongation factors SPT6 and PAF19, 22. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 3 SPT6 is both a transcription elongation factor and a histone chaperone, playing an essential role in transcription processivity, elongation across nucleosome barriers and nucleosome positioning23, 24, 25, 26. Here we report the cryo-EM structure of a transcribing Pol II-DSIF- SPT6-U1 snRNP complex. Additionally, we identify that U1 snRNP is recruited to the elongation complex through multiple interaction sites, enabling efficient co-transcriptional spliceosome assembly.

Human U1 snRNP consists of U1 snRNA, seven Sm proteins and three U1-specific proteins, U1-70K, U1A and U1C. Both U1-70K and U1A contain RNA recognition motifs (RRMs) and associate with the U1 snRNA stem loops, while U1C comprises a zinc finger domain that facilitates 5´ SS recognition27 (Fig. 1a). We first investigated the effect of P- TEFb phosphorylation on the Pol II-U1 snRNP interaction. We generated a CRISPR-knockin cell line with an N-terminal TwinStrep tag on the largest subunit of Pol II, RPB1, and purified human Pol II (Methods). Using the TwinStrep tag on Pol II, we performed pulldown experiments in the absence of any nucleic acids to capture solely protein-mediated interactions (Fig. 1b). Consistent with the Pol II-U1 snRNP structure13, U1 snRNP bound directly to Pol II and its binding increased with Pol II phosphorylation by P-TEFb (Fig. 1b compare lanes 6 and 9). This enhanced binding is attributed to the specific interaction of U1 snRNP with the P-TEFb phosphorylated Pol II CTD, but not with the non-phosphorylated CTD (Extended Data Fig. 1). Pulldown experiments using U1-specific proteins revealed that the phosphorylated Pol II CTD bound specifically to the U1-70KRRM domain (Extended Data Fig. 1, lane 10), the same domain that mediates the direct Pol II-U1 snRNP interaction13. We used the RRM domain of U1-70K in the pulldown as full-length U1-70K is insoluble. To assess the effect of elongation factors on the Pol II-U1 snRNP interaction, we added DSIF and SPT6 to Pol II and treated them with P-TEFb phosphorylation before incubating .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 4 with U1 snRNP. In the absence of nucleic acids, DSIF failed to associate with Pol II, whereas SPT6 formed a stochiometric complex with Pol II (Fig. 1b). Interestingly, association of U1 snRNP to phosphorylated Pol II is further enhanced by SPT6 (Fig. 1b compare lanes 9 and 12), suggesting SPT6 may play a role in U1 snRNP recruitment and thereby facilitating co- transcriptional splicing. To gain molecular insights into U1 snRNP recruitment during transcription elongation, we assembled a mammalian transcription elongation complex (EC) with human DSIF and SPT6 on a DNA-RNA scaffold. The scaffold contains a DNA mismatch bubble that enables formation of a 9-base pair DNA-RNA hybrid duplex, and a modified MINX pre-mRNA that comprises a 5´ exon and a truncated intron of 34 nucleotides (Extended Data Fig. 2a). The assembled EC-DSIF-SPT6 complex was phosphorylated by P-TEFb and purified by size exclusion chromatography before its incubation with purified human U1 snRNP and subjected to single particle cryo-EM analysis (Extended Data Fig. 2-4). Although PAF does not interfere with U1 snRNP binding to Pol II in vitro13, we omitted PAF in the assembly because it further increased the flexibility of an already highly mobile complex. We obtained a cryo-EM reconstruction of the EC-DSIF-SPT6-U1 snRNP complex at an overall resolution of 3.5 Å (Fig. 1c, Extended Data Fig. 2-4, Extended Data Table 1, Supplementary Video 1). Both U1 snRNP and SPT6 are highly mobile on the Pol II surface. Focused refined maps with a local resolution of 7.3 Å for U1 snRNP and 6.2 Å for SPT6 allowed confident docking of previous models13, 28 (Extended Data Fig. 4). While both SPT6 and DSIF engage the stalk domain of Pol II, DSIF additionally interacts with the upstream DNA, possibly explaining the lack of DSIF binding to Pol II in the absence of nucleic acids in our pulldown experiment (Fig. 1b). The direct Pol II-U1 snRNP interface is mediated by the RRM domain of U1-70K which contacts the RPB2 and RPB12 subunits of Pol II, consistent with the Pol II-U1 snRNP structure13. Therefore, the U1 snRNP interaction with .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 5 Pol II is compatible with the presence of elongation factors DSIF and SPT6 on the Pol II surface. However, we could not resolve the interactions of U1 snRNP with SPT6 and the phosphorylated Pol II CTD in our cryo-EM reconstruction, likely due to the mobile nature of the complex. To understand the role of SPT6 in U1 snRNP recruitment, we performed pulldown experiments using an MBP-tagged SPT6 and observed direct binding of U1 snRNP to SPT6 regardless of its phosphorylation state (Fig. 2a, b). Human SPT6 contains a core domain that engages the Pol II stalk domain (Fig. 1c) and a tSH2 domain that interacts with the phosphorylated Pol II CTD linker region19, 22. The SPT6 core and tSH2 domains are flanked by an N-terminal domain (NTD) and a C-terminal region (CTR) that are only present in eukaryotes and are less well characterized (Fig. 2a). SPT6NTD, a highly acidic and largely unstructured region, is important for its interaction with histones, while SPT6CTR is mostly disordered. Deletion of the CTR (SPT6ΔCTR) completely abolished U1 snRNP binding (Fig. 2b lane 10), whereas eliminating either the core and tSH2 domains (SPT6NTD+CTR) or the NTD (SPT6ΔNTD) did not affect U1 snRNP association (Fig. 2c lanes 6 and 10). Consistently, while SPT6CTR itself is able to bind U1 snRNP (Fig. 2b lane 12), neither SPT6Core nor SPT6NTD interacts with U1 snRNP (Fig. 2b, c lanes 8), underlying the importance of SPT6CTR in mediating the SPT6-U1 snRNP interaction. We next investigated which U1-specific proteins are responsible for interacting with SPT6CTR and found that SPT6CTR bound specifically to U1A, but not U1C and U1-70KRRM (Fig. 2d). We further tested U1A interaction with individual SPT6 truncations (Fig. 2a, e). Consistent with U1 snRNP, deletion of the CTR (SPT6ΔCTR) eliminated binding of U1A, while the CTR is the only domain capable to engage U1A by itself (Fig. 2e lanes 10 and 8). This data was further supported by isothermal titration calorimetry (ITC) experiments, .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 6 obtaining a dissociation constant (Kd) of 1.96 µM for U1A interaction with wildtype SPT6 and a Kd of 4.0 µM for U1A interaction with the SPT6CTR (Fig. 2a, Extended Data Fig. 5). U1A associates with stem loop 2 of U1 snRNA, located distant from the Pol II-U1 snRNP interface (Fig. 1c). The dynamic nature of both U1 snRNP and SPT6 on Pol II makes it challenging to resolve their interactions by cryo-EM. Additionally, AlphaFold3 (ref29) was unable to generate confident predictions when full-length SPT6 and U1A were used. However, guided by the SPT6CTR-U1A interaction from our pulldown and ITC experiments, AlphaFold3 produced a prediction for the SPT6CTR bound to U1A with high confidence (Fig. 2f). A C-terminal ⍺-helix of SPT6 (residues 1671-1696) harbouring aromatic and hydrophobic residues contacts a hydrophobic surface of the N-terminal RRM domain of U1A located on the opposite side of its U1 snRNA interface. Taken together, our results reveal that U1 snRNP is recruited to the transcription elongation complex through three contact points: while the Pol II body and the phosphorylated Pol II CTD interact directly with U1-70K, the elongation factor SPT6 associates with U1A (Fig. 2g, Supplementary Video 1). All three interactions are mediated by the RRM domains in U1 snRNP. This multivalent interaction facilitates the swift recruitment of U1 snRNP to the nascent transcripts, thereby allowing efficient and accurate co- transcriptional spliceosome assembly. Furthermore, U1 snRNP is involved in preventing pre- mature transcription termination independent of splicing30 and increases Pol II elongation rates to allow synthesis of long genes31. Therefore, the observed direct interactions of Pol II and SPT6 with U1 snRNP likely also ensure transcription processivity during elongation. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 7 Main references 1. Kotovic KM, Lockshon D, Boric L, Neugebauer KM. Cotranscriptional recruitment of the U1 snRNP to intron-containing genes in yeast. Mol Cell Biol 23, 5768-5779 (2003). 2. Lacadie SA, Rosbash M. Cotranscriptional spliceosome assembly dynamics and the role of U1 snRNA:5'ss base pairing in yeast. Mol Cell 19, 65-75 (2005). 3. Listerman I, Sapra AK, Neugebauer KM. Cotranscriptional coupling of splicing factor recruitment and precursor messenger RNA splicing in mammalian cells. Nat Struct Mol Biol 13, 815-822 (2006). 4. Wallace EWJ, Beggs JD. Extremely fast and incredibly close: cotranscriptional splicing in budding yeast. RNA 23, 601-610 (2017). 5. Bird G, Zorio DA, Bentley DL. RNA polymerase II carboxy -terminal domain phosphorylation is required for cotranscriptional pre -mRNA splicing and 3' -end formation. Mol Cell Biol 24, 8963-8969 (2004). 6. Sakharkar MK, Chow VT, Kangueane P. Distributions of exons and introns in the human genome. In Silico Biol 4, 387-393 (2004). 7. Yuryev A, et al. The C-terminal domain of the largest subunit of RNA polymerase II interacts with a novel set of serine/arginine-rich proteins. Proc Natl Acad Sci U S A 93, 6975-6980 (1996). 8. Nojima T, et al. Mammalian NET-Seq Reveals Genome -wide Nascent Transcription Coupled to RNA Processing. Cell 161, 526-540 (2015). 9. Misteli T, Spector DL. RNA polymerase II targets pre -mRNA splicing factors to transcription sites in vivo. Mol Cell 3, 697-705 (1999). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 8 10. Harlen KM, Trotta KL, Smith EE, Mosaheb MM, Fuchs SM, Churchman LS. Comprehensive RNA Polymerase II Interactomes Reveal Distinct and Varied Roles for Each Phospho-CTD Residue. Cell Rep 15, 2147-2158 (2016). 11. David CJ, Boyne AR, Millhouse SR, Manley JL. The RNA polymerase II C -terminal domain promotes splicing activation through recruitment of a U2AF65-Prp19 complex. Genes Dev 25, 972-983 (2011). 12. Natalizio BJ, Robson-Dixon ND, Garcia-Blanco MA. The Carboxyl-terminal Domain of RNA Polymerase II Is Not Sufficient to Enhance the Efficiency of Pre -mRNA Capping or Splicing in the Context of a Different Polymerase. J Biol Chem 284, 8692- 8702 (2009). 13. Zhang S, Aibara S, Vos SM, Agafonov DE, Luhrmann R, Cramer P. Structure of a transcribing RNA polymerase II-U1 snRNP complex. Science 371, 305-309 (2021). 14. Mount SM, Pettersson I, Hinterberger M, Karmas A, Steitz JA. The U1 small nuclear RNA-protein complex selectively binds a 5' splice site in vitro. Cell 33, 509-518 (1983). 15. Leader Y , et al. The upstream 5' splice site remains associated to the transcription machinery during intron synthesis. Nat Commun 12, 4545 (2021). 16. Wei P, Garber ME, Fang SM, Fischer WH, Jones KA. A novel CDK9 -associated C- type cyclin interacts directly with HIV -1 Tat and mediates its high -affinity, loop - specific binding to TAR RNA. Cell 92, 451-462 (1998). 17. Marshall NF, Price DH. Purification of P-TEFb, a transcription factor required for the transition into productive elongation. J Biol Chem 270, 12335-12338 (1995). 18. Wada T, Takagi T, Yamaguchi Y, Watanabe D, Handa H. Evidence that P -TEFb alleviates the negative effect of DSIF on RNA polymerase II -dependent transcription in vitro. EMBO J 17, 7395-7403 (1998). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 9 19. Vos SM, Farnung L, Urlaub H, Cramer P. Structure of paused transcription complex Pol II-DSIF-NELF. Nature 560, 601-606 (2018). 20. Ping YH, Rana TM. Tat -associated kinase (P -TEFb): a component of transcription preinitiation and elongation complexes. J Biol Chem 274, 7399-7404 (1999). 21. Fujinaga K, Irwin D, Huang Y, Taube R, Kurosu T, Peterlin BM. Dynamics of human immunodeficiency virus transcription: P -TEFb phosphorylates RD and dissociates negative effectors from the transactivation response element. Mol Cell Biol 24, 787- 795 (2004). 22. Vos SM, et al. Structure of activated transcription complex Pol II -DSIF-PAF-SPT6. Nature 560, 607-612 (2018). 23. Bortvin A, Winston F. Evidence that Spt6p controls chromatin structure by a direct interaction with histones. Science 272, 1473-1476 (1996). 24. Kaplan CD, Morris JR, Wu C, Winston F. Spt5 and spt6 are associated with active transcription and have characteristics of general elongation factors in D. melanogaster. Genes Dev 14, 2623-2634 (2000). 25. Endoh M, et al. Human Spt6 stimulates transcription elongation by RNA polymerase II in vitro. Mol Cell Biol 24, 3324-3336 (2004). 26. Ardehali MB, et al. Spt6 enhances the elongation rate of RNA polymerase II in vivo. EMBO J 28, 1067-1077 (2009). 27. Kondo Y, Oubridge C, van Roon AM, Nagai K. Crystal structure of human U1 snRNP, a small nuclear ribonucleoprotein particle, reveals the mechanism of 5' splice site recognition. Elife 4, (2015). 28. Zhang L, Gordiyenko Y, Morgan T, Franco C, Vidaković AT, Zhang S. Structural basis of RECQL5 -induced RNA polymerase II transcription braking and subsequent reactivation. bioRxiv, 2025.2001.2029.635449 (2025). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 10 29. Abramson J , et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 630, 493-500 (2024). 30. Kaida D , et al. U1 snRNP protects pre -mRNAs from premature cleavage and polyadenylation. Nature 468, 664-668 (2010). 31. Mimoso CA, Adelman K. U1 snRNP increases RNA Pol II elongation rate to enable synthesis of long genes. Mol Cell 83, 1264-1279 e1210 (2023). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 11 Fig. 1 | Structure of the EC-DSIF-SPT6-U1 snRNP complex. a, Domain organization of U1A, U1-70K and U1C. RRM: RNA recognition motif, ZnF: zinc finger. b, Pulldown of U1 snRNP using TwinStrep-tagged human Pol II in different phosphorylation states, along with elongation factors DSIF and SPT6. The prey protein U1-70K as a part of U1 snRNP is highlighted in a purple rectangle. The pulldown was performed in the absence of any nucleic acids and repeated in triplicate. pPol II: phosphorylated Pol II, pRPB1: phosphorylated RPB1, IN: input, FT: flow-through, E: elution. Asterisk indicates RPB1 degradation. c, Structure of the EC-DSIF-SPT6-U1 snRNP complex in top view. Pol II is shown in light grey surface representation except for RPB2 in yellow and RPB12 in lime. Elongation factors SPT6 (blue) and DSIF (green) as well as U1 snRNP are shown in cartoon representation. The backbone of U1 snRNA is coloured in lavender, U1-70K in dark purple, U1A in violet and Sm proteins in light pink. Template DNA is coloured dark blue, non-template DNA in cyan and RNA in red. SL: stem loop. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 12 Fig. 2 | SPT6 interacts directly with U1 snRNP. a, Domain organization of SPT6 and truncated SPT6 constructs. A summary of U1 snRNP and U1A binding to all SPT6 constructs in pulldown assays are shown on the right, together with the dissociation constant (Kd) and number of binding sites (N) determined by ITC for the SPT6-U1A interaction. b, c, Pulldown of U1 snRNP using different MBP-tagged SPT6 constructs. The prey protein U1-70K as a part of U1 snRNP is highlighted in a purple rectangle. MBP is used as a negative control. IN: input, E: elution. d, Pulldown of U1-specific proteins using MBP-tagged SPT6CTR. The prey proteins U1A, U1C and U1-70KRRM are highlighted in rectangles. e, Pulldown of U1A using different MBP-tagged SPT6 constructs. The prey protein U1A is highlighted in a violet rectangle. All pulldowns were repeated in triplicate. f. AlphaFold3 prediction of the .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 13 SPT6CTR-U1A interaction, superimposed onto the U1 snRNP structure13. The predicted aligned error plot is shown on the right, with the interface highlighted with red boxes. g, Cartoon schematic showing the three contact points between U1 snRNP and the transcription elongation complex: U1-70KRRM with RPB2 and RPB12, U1-70KRRM with the phosphorylated Pol II CTD, and U1A with SPT6CTR. Arrows indicate the direction of transcription. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 14 Extended Data Fig. 1 | U1 snRNP interacts directly with the phosphorylated Pol II CTD. Pulldown of U1 snRNP (U1) and U1-specific proteins with MBP-tagged human Pol II CTD. U1 snRNP interacts specifically with the P-TEFb phosphorylated Pol II CTD. P-TEFb consists of CDK9 and cyclin T1, a truncated cyclin T1 (1-272) was used. The prey proteins U1-70KRRM, U1A and U1C are highlighted in coloured rectangles. Asterisk indicates MBP- contamination from the U1C preparation that was enriched in the elution. This pulldown was repeated in triplicate. Extended Data Fig. 2 | Preparation of the EC-DSIF-SPT6-U1 snRNP complex. a, Nucleic acid scaffold used for cryo-EM, with template DNA in dark blue, non-template DNA in cyan and RNA in red. b, Assembled EC-DSIF-SPT6-U1 snRNP complex on a 4-12% NuPAGE Bis-Tris gel, run in MOPS, stained with Instant Blue. c, Representative micrograph of the EC-DSIF-SPT6-U1 snRNP complex collected on the 300 kV Titan Krios with the K3 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 15 detector in electron counting mode. d, Two-dimensional averages of the EC-DSIF-SPT6-U1 snRNP complex. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 16 Extended Data Fig. 3 | Cryo-EM data processing of the EC-DSIF-SPT6-U1 snRNP complex. 2D-classification, followed by 3D-refinement and local 3D-classification were performed to remove bad particles from the datasets (a). Signal subtraction followed by focused 3D-classification without alignment was performed using a soft mask on U1 snRNP (b, c). Particles containing densities of U1 snRNP were reverted to original particles and 3D- refined, resulting in a reconstruction of 3.6 Å resolution (d), with a locally refined map at 6.7 Å for U1 snRNP (e). Following signal subtraction and focused 3D-classification without alignment with a soft mask on SPT6 (f), particles containing densities of SPT6 were reverted to original particles and 3D-refined. This resulted in a final reconstruction of EC-DSIF-SPT6- U1 snRNP at an overall resolution of 3.5 Å (g). Particle subtraction followed by local refinement improved the local resolution of U1 snRNP and SPT6 (h, i). The cryo-EM densities were coloured according to Fig. 1c. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 17 Extended Data Fig. 4 | Cryo-EM densities, local resolution estimation and FSC curves of the EC-DSIF-SPT6-U1 snRNP complex. a-i, Gold-standard Fourier Shell Correlation (FSC) curves, local resolution estimations and cryo-EM densities with fitted models for the overall map (a-c), U1 snRNP (d-f) and SPT6 (g-i). The orange box in e and f indicate the Pol II-U1 snRNP interface. The locally filtered and sharpened map was shown for the overall map, and sharpened maps were shown for the focused refined maps of U1 snRNP and SPT6. j, Model versus map FSC for the locally filtered and sharpened overall map using the FSC standard of 0.5. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 18 Extended Data Fig. 5 | ITC analysis for the interaction between U1A and SPT6. a-g, Isothermal titration calorimetry (ITC) thermograms of U1A binding to SPT6 (a), SPT6NTD+CTR (b), SPT6ΔNTD (c), His6-MBP-tagged SPT6CTR (d), SPT6NTD (e), SPT6ΔCTR (f) and MBP as a negative control (g). The upper panels show a representative thermogram. The lower panels show the integrated heat changes and fitting of the data. h, Summary of the dissociation constant (Kd) and stoichiometry (N) for U1A-SPT6 interactions. Error bars represent standard deviation from triplicates. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 19 Extended Data Table 1 | Statistics of cryo-EM reconstructions and structural model. Overall (EMD-53087) SPT6 (EMD-53088) U1 snRNP (EMD-53089) Data collection and processing Microscope Titan Krios Voltage (kV) 300 Camera K3 Magnification 81 000 x Pixel size (Å/pixel) 1.05 Electron exposure (e-/Å2) 40 Exposure rate (e-/Å2/frame) 1.01 Number of frames per movie 40 Defocus range (μm) 0.5-2.0 Automation software SerialEM Symmetry imposed C1 Initial particle numbers 2,039,129 Final particle numbers 52,065 Map sharpening B factor (Å2) -111.322 -500 -513.8 Map resolution (Å, FSC=0.143) 3.5 6.2 7.3 Refinement Initial models used (PDB) 7B0Y, 9HVQ Model resolution (Å) 3.7 Model composition Non-hydrogen atoms 56,145 Protein residues 6,297 Nucleic acid residues 263 Ligands Zn:8 Mg:1 B factors (Å2) Protein 327.52 Nucleotide 554.97 Ligand 239.11 R.m.s. deviations Bond lengths (Å) 0.004 Bond angles (°) 0.568 Validation MolProbity score 1.57 Clash score 5.98 Poor rotamers (%) 0.00 Cß deviations (%) 0 Ramachandran plot Favored (%) 96.35 Allowed (%) 3.65 Outliers (%) 0.00 PDB code 9QEQ Supplementary Video 1 | The transcription elongation complex recruits U1 snRNP through multiple interaction sites. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 20

Cloning and protein expression Human U1A, U1C and U1-70KRRM were amplified from cDNA isolated from FreeStyle293 cells (Thermo Fisher) and cloned into 1C vector (addgene no. 29654) with an N-terminal His6-MBP-TEV tag using ligation independent cloning32. All SPT6 truncations were generated using Quikchange33 or NEB HiFi assembly in 438C vector (addgene no. 55220) with an N-terminal His6-MBP-TEV tag. Proteins cloned into 438C vector were expressed in High Five cells (Gibco). 1 L High Five cells in Sf-900TM II SFM medium (Gibco) were infected with P2 virus and grown for 50 to 72 hours. Cells were harvested by centrifugation at 1000xg for 18 min and frozen in liquid nitrogen and stored at -80 °C before protein purification. Proteins cloned into 1C vector were expressed in BL21 (DE3) RIL cells (Agilent) in LB medium. Expression was induced with 1 mM IPTG when the cell density reached an OD of 0.6-0.8. Proteins were expressed overnight at 18 °C before harvesting and storage at -80 °C. Cell line genome editing The HEK293 cell line was CRISPR/Cas9 edited to incorporate a TwinStrep tag followed by a TEV cleavage site at the N-terminus of RPB1. Cloning and endogenous knock- in utilized a MMEJ-assisted gene knock-in strategy34 as described previously35 with minor modifications. In brief, HEK293 cells were plated 24 h before the start of the experiment to ensure exponential growth. For knock-in, 1 million cells per reaction were transfected with 2 μg DNA (microhomology-containing repair template plasmid and sgRNA-Cas9 expression vector) using Amaxa Nucleofector with SF Cell Line 4D-Nucleofector X Kit L (Lonza V4XC-2012) and program CM130. After electroporation, cells were taken up in 500 μl pre- .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 21 warmed growth medium, transferred to 6-well plates containing 1.5 mL warm growth medium and allowed to recover for a total of four days at 37 °C and subcultured to maintain the cells as needed. On day four, antibiotic selection was initiated by treating with 1 μg/ml puromycin for five days. Presence of the insert in the pool was validated by performing genotyping PCR before proceeding to single cell cloning. Individual clones were isolated by performing serial dilution in 96-well plates and expanding clones for 12-14 days. Obtained clones were characterized by performing genotyping PCR and Sanger sequencing of PCR amplicons of the genomic integration site. The following primer and insert sequences were used: sgRNA target sequence: gcctccgccatgcacggggg HR template cloning forward primer: gcgttacatagcatcgtacgcgtacgtgtttggcctgcctccgccatgcacgggaccgagtacaagcccacg HR template cloning reverse primer: agcattctagagcatcgtacgcgtacgtgtttggcccccgagggggggccacccccatgtccggacgcgtttgactgg Puro_His10_TwinStrep sequence: atgaccgagtacaagcccacggtgcgcctcgccacccgcgacgacgtccccagggccgtacgcaccctcgccgccgcgttcgccg actaccccgccacgcgccacaccgtcgatccggaccgccacatcgagcgggtcaccgagctgcaagaactcttcctcacgcgcgtc gggctcgacatcggcaaggtgtgggtcgcggacgacggcgccgcggtggcggtctggaccacgccggagagcgtcgaagcggg ggcggtgttcgccgagatcggcccgcgcatggccgagttgagcggttcccggctggccgcgcagcaacagatggaaggcctcctg gcgccgcaccggcccaaggagcccgcgtggttcctggccaccgtcggcgtctcgcccgaccaccagggcaagggtctgggcagc gccgtcgtgctccccggagtggaggcggccgagcgcgccggggtgcccgccttcctggagacctccgcgccccgcaacctcccct tctacgagcggctcggcttcaccgtcaccgccgacgtcgaggtgcccgaaggaccgcgcacctggtgcatgacccgcaagcccgg tgccggctctggagctactaacttcagcctgctgaagcaggctggagacgtggaggagaaccctggacctcatcatcatcaccaccat .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 22 caccatcatcacggttcttcttggtctcacccccaatttgagaaaggcggtggcagcggcggcggtagcggaggtggcagctggtca cacccacaattcgagaaaggcgccgagaatctttatttccagtcaaacgcgtccgga Genotyping forward primer: gtagtgaggtttgcgcctgc Genotyping reverse primer: gaaagcgccaagtttccgca Protein purification The porcine Pol II was purified from Sus scrofa domesticus thymus and human U1 snRNP from HeLa cells as described13. Porcine Pol II has a 99.9% sequence identity to human Pol II. Human P-TEFb, DSIF and SPT6 were purified as described22, while His6- MBP-tagged human Pol II CTD was purified as described28. For the purification of TwinStrep-tagged human Pol II, CRISPR/Cas9 edited HEK293 cells were grown in suspension in Expi293 medium (Thermo Fisher) and harvested at a density between 5-6 million/ml. The cell pellet was resuspended in 0 M buffer (50 mM Tris- HCl pH 7.9, 5 mM MgCl2, 0.5 mM EDTA, 10 % Glycerol, 2 mM DTT, 1 mM Na2S2O5, 1 mM PMSF) supplemented with EDTA-free protease inhibitor tablets (Roche). The resuspended cells were sonicated before slowly adding an equal volume of 0.6 M ammonium sulfate buffer (0 M buffer with 0.6 M ammonium sulfate) supplemented with EDTA-free protease inhibitor tablets (Roche) and 0.01 mg/ml DNase (Sigma-Aldrich), which was further sonicated. The mixture was cleared by centrifugation and the supernatant was precipitated with 50% saturated ammonium sulfate for 1 h. The pellet was re-dissolved in 0 M buffer until the conductivity of the sample matched that of the 0.5 M ammonium sulfate buffer, cleared by centrifugation and applied onto the Strep-TactinXT 4Flow column (IBA). The column was washed with 0.5 M buffer, followed by 0.18 M buffer, and eluted with 0.18 M buffer supplemented with 50 mM biotin. The eluate was loaded onto an UNO Q1 column (BioRad), washed with 0.18 M buffer and eluted with a gradient of 0.5 M buffer. The peak fractions .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 23 were concentrated using an Amicon Ultra Centrifugal Filter with 100 kDa molecular weight cutoff (MWCO) and buffer exchanged to Pol II storage buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 10 μM ZnCl2, 2 mM DTT). For all His6-MBP-tagged human SPT6 constructs, cell pellets were resuspended in SPT6 Lysis Buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 30 mM Imidazole, 5% glycerol, 1 mM DTT) supplemented with 1 mM PMSF and EDTA-free protease inhibitor tablets (Roche), sonicated, and clarified by centrifugation. Cleared lysates were applied onto the HisTrap HP column (Cytiva), washed with SPT6 Lysis Buffer and Buffer B (50 mM Tris- HCl pH 7.5, 1 M NaCl, 5% glycerol, 1 mM DTT), followed by elution with SPT6 Lysis Buffer supplemented with 300 mM Imidazole. The eluate was applied onto a home-packed amylose column (NEB), washed with SPT6 Lysis Buffer and eluted with 50 mM Tris-HCl pH 7.5, 300 mM NaCl, 50 mM maltose, 5% glycerol, 1 mM DTT. The eluate was diluted to 75 mM NaCl with Buffer A (50 mM Tris-HCl pH 7.5, 5% glycerol, 1 mM DTT) and applied onto a HiTrap Heparin HP column (Cytiva). The protein was eluted with a gradient of Buffer A and Buffer B, followed by a final size exclusion step on HiLoad 16/600 Superdex 75 pg or 200 pg columns (Cytiva) in SPT6 SEC Buffer (20 mM HEPES-NaOH pH 7.5, 300 mM NaCl, 5% glycerol, 1 mM DTT). Peak fractions were concentrated, frozen in liquid nitrogen and stored at -80 °C. For U1A, U1C and U1-70KRRM, cell pellets were resuspended in U1A Lysis Buffer (20 mM HEPES-NaOH pH 7.5, 500 mM NaCl, 10% glycerol, 1 mM DTT) supplemented with 1 mM PMSF and EDTA-free protease inhibitor tablets (Roche). Resuspended cells were sonicated, cleared by centrifugation, and applied onto the HisTrap HP column. The column was washed with U1A Lysis Buffer followed by Buffer D (20 mM HEPES-NaOH pH 7.5, 1 M NaCl, 10% glycerol, 1 mM DTT) and eluted with U1A Lysis Buffer supplemented with 300 mM Imidazole. The eluate was loaded onto a home-packed amylose column, washed .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 24 with U1A Lysis Buffer and eluted with 20 mM HEPES-NaOH pH 7.5, 300 mM NaCl, 100 mM maltose, 10% glycerol and 1 mM DTT. The eluate was diluted to 100 mM NaCl with Buffer C (20 mM HEPES-NaOH pH 7.5, 10% glycerol and 1 mM DTT) and digested with TEV protease overnight at 4 °C to cleave off the His6-MBP tag. Proteins were applied onto a HiTrap Heparin HP column and eluted with a gradient of Buffer C and Buffer D. Peak fractions containing digested proteins were loaded onto a HiLoad 16/600 Superdex 75 pg column in U1A SEC Buffer (20 mM HEPES-NaOH pH 7.5, 300 mM NaCl, 1 mM DTT). Peak fractions were concentrated, frozen in liquid nitrogen and stored at -80 °C. Pulldown assay For the pulldown of U1 snRNP with Pol II, TwinStrep-tagged human Pol II (12 pmol) alone or with SPT6 (36 pmol) and DSIF (36 pmol) was in vitro phosphorylated with P-TEFb (3 pmol) and ATP (1 mM) in SEC100 buffer (20 mM HEPES-NaOH pH 7.5, 100 mM NaCl, 3 mM MgCl2 and 1 mM DTT) for 30 min at 30 °C. P-TEFb storage buffer (20 mM HEPES pH 7.5, 300 mM NaCl, 10% glycerol, 1 mM DTT) is used instead of P-TEFb as a negative control for the non-phosphorylated Pol II condition. Samples were incubated with U1 snRNP (36 pmol) on ice for 30 min, followed by incubation with the StrepTactinXT 4Flow high- capacity resin (IBA) equilibrated in K75 Buffer (20 mM HEPES-NaOH pH 7.5, 75 mM KCl, 3 mM MgCl2 and 1 mM DTT) at 4 °C for 1 hour. The resin was washed with K75 Buffer and eluted with K75 Buffer supplemented with 50 mM biotin. The eluate was separated on a 4- 12% NuPAGE Bis-Tris gel and stained with InstantBlue (Abcam). For the pulldown of U1 snRNP or U1-specific proteins with Pol II CTD, His6-MBP- tagged Pol II CTD (50 pmol) was in vitro phosphorylated with P-TEFb (12.5 pmol) and ATP (1 mM) in SEC100 Buffer for 30 min at 30 °C. P-TEFb storage buffer was used instead of P- TEFb as a negative control for the non-phosphorylated CTD condition. The CTD was incubated with U1 snRNP (50 pmol), U1A (100 pmol), U1C (100 pmol) or U1-70KRRM (100 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 25 pmol) on ice for 30 min, followed by incubation with amylose resin equilibrated in G- SEC150 Buffer (20 mM HEPES-NaOH pH 7.5, 150 mM NaCl, 3 mM MgCl2, 5% glycerol and 1 mM DTT) at 4 °C for 1 hour. The resin was washed with G-SEC150 Buffer and eluted with G-SEC150 Buffer supplemented with 20 mM maltose. For the pulldown of U1 snRNP with different SPT6 constructs, equimolar U1 snRNP (30 pmol) was incubated with His6-MBP-tagged SPT6 proteins (30 pmol) at room temperature for 30 min in SEC100 Buffer in Fig. 2b or SEC200 Buffer (SEC100 with 200 mM NaCl) in Fig. 2c based on the stability of the SPT6 constructs, followed by incubation with the amylose resin (NEB) at 4 °C for 1 hour. The resin was washed with SEC100/200 Buffer and eluted with SEC100/200 Buffer supplemented with 20 mM maltose. Pulldown of U1A with different SPT6 constructs was performed as above in SEC200 buffer except for 3x molar excess of U1A (300 pmol) was incubated with His6-MBP-tagged SPT6 constructs (100 pmol). For the pulldown of SPT6CTR with U1-specific proteins, His6- MBP-tagged SPT6CTR (200 pmol) was incubated with 4x molar excess of U1A, U1C, U1- 70KRRM (800 pmol). The pulldown was performed as above in SEC100 buffer. ITC Affinities between SPT6 constructs and U1A were determined by ITC at 25 °C with a Malvern Panalytical ITC200 instrument in SEC300 Buffer (20 mM HEPES-NaOH pH 7.5, 300 mM NaCl). In a typical ITC experiment, U1A was loaded into a 40 μl syringe at a concentration between 150-550 μM and the SPT6 construct was placed into the sample cell at a concentration between 10-35 μM. Titrations consisted of 19 injections of 2 μL preceded by a small 0.5 μL pre-injection that was not used during curve fitting. Experiments were performed at a reference power of 6 μcal/s and an initial delay of 180 s with injections at 180 s intervals with constant stirring at 750 rpm. Control measurements of injections of protein .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 26 into buffer were performed and these control heats were close to the values seen for buffer into buffer blank runs. All ITC binding data were corrected with the appropriate control heats of dilution and fitted using the one set of binding sites’ model in Malvern Panalytical PEAQ- ITC analysis software (v1.41). Experiments were performed at least 3 times with different batches of proteins. RNA preparation The following RNA construct was generated for this study: 5’-CUU GGA UCG GAA ACC CGU CGG CCU CCG ACA GGU AAG UAU AUG UAU AAC CGG AGA GGG AAC CCA CU-3’ The sequence that is complementary to U1 snRNA is in bold and the sequence that is covered within Pol II is in italic. The construct was cloned into the pUC18 vector with a T7 promotor at the 5’ end and a hepatitis delta virus ribozyme at the 3’ end. RNA was prepared by T7 transcription, followed by purification by gel extraction and capped with Vaccinia capping enzyme as described13. Sample preparation for cryo-EM The EC-DSIF-SPT6-U1 snRNP complex was formed on a DNA scaffold with the following sequences: template DNA 5′-GCT CCC AGC TCC CTG CTG GCT CCG AGT GGG TTC TGC CGC TCT CAA TGG-3′, non-template DNA 5′-CCA TTG AGA GCG GCC CTT GTG TTC AGG AGC CAG CAG GGA GCT GGG AGC-3′. The DNA scaffold contains 14 nucleotides of upstream DNA, 23 nucleotides of downstream DNA and a mismatch bubble of 11 nucleotides, of which 9 nucleotides of the template DNA base-pairs with the RNA. The DNA oligos were synthesized by IDT and dissolved in water. RNA and template DNA were mixed in equimolar ratio (30 µM) in 20 mM HEPES- NaOH pH 7.5, 100 mM NaCl and 3 mM MgCl2, and annealed by heating up at 60 °C for 4 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 27 min followed by decreasing the temperature by 1 °C min-1 steps to a final temperature of 30 °C in a thermocycler. S. scrofa Pol II (50 pmol) and the RNA–template DNA hybrid (100 pmol) were incubated for 10 min at 30 °C, followed by addition of the non-template DNA (200 pmol) and incubation at 30 °C for another 10 min. After adding DSIF and SPT6 (150 pmol each), phosphorylation reaction was carried out with 0.4 µM P-TEFb and 1 mM ATP in SEC100 buffer for 30 min at 30 °C. The assembled EC-DSIF-SPT6 complex was applied onto a Superose 6 Increase 3.2/300 column (GE Healthcare) equilibrated with SEC100 buffer containing 0.5 mM tris(2-carboxyethyl)phosphine (TCEP) instead of DTT. U1 snRNP was further purified on the Superose 6 Increase 3.2/300 column (GE Healthcare) in SEC100 buffer. The peak fraction of the EC-DSIF-SPT6 complex was mixed with the peak fraction of U1 snRNP at a molar ratio of 1:1.5, incubated on ice for 30 min, crosslinked with 0.05% glutaraldehyde on ice for 45 min and used directly for freezing grids. Similar results were obtained when combining the EC-DSIF-SPT6 complex and U1 snRNP before the size exclusion purification step. Samples were diluted to a concentration of ~150 nM and 2 µl was applied to each side of the R2/2 UltrAuFoil grids (Quantifoil) that had been glow-discharged for 100 s. After incubation of 10 s and blotting for 4 s, the grid was vitrified by plunging it into liquid ethane with a Vitrobot Mark IV (FEI Company) operated at 4 °C and 100% humidity. Cryo-EM data collection and processing Cryo-EM data were collected on the 300 kV Titan Krios (Thermo Fisher) with a K3 summit direct detector (Gatan) and a GIF quantum energy filter (Gatan) operated with a slit width of 20 eV. Automated data collection was performed with SerialEM36 at a nominal magnification of 81,000x, corresponding to a pixel size of 1.05 Å/pixel. Image stacks of 40 movie frames were collected with a defocus range of -0.5 to -2.0 µm in electron counting mode and a dose rate of 1.01 e-/Å2/frame. A total of 18830 image stacks were collected. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 28 All movie frames were aligned and the contrast transfer function (CTF) parameters were calculated in Warp37. Particles in 380 pixels x 380 pixels were selected by automatic particle picking in Warp. The following steps were performed in RELION 5.0 (ref38, 39) to exclude bad particles from the dataset: 1) Two-dimensional (2D) classification was performed and particles in bad classes with poorly recognizable features were excluded. 2) In the second round of 2D classification, free U1 snRNP particles were excluded, leaving only Pol II containing particles. 3) The remaining particles were refined using three-dimensional (3D) refinement with a soft mask on Pol II and divided into six classes using 3D classification in RELION with local fine-angle search (0.9 degree). All 3D classes with bad particles were discarded. All good particles with Pol II were combined and 3D refined using a soft mask on Pol II. To separate Pol II alone particles from the U1 snRNP containing particles, signal subtraction followed by focused 3D classification of the subtracted particles without alignment was performed using a large mask near the RNA exit site. The classes with an extra density corresponding to U1 snRNP were combined (12.4%), reverted to original particles and 3D refined with a soft mask on Pol II. Subsequently, the SPT6-containing particles were selected using particle subtraction followed by 3D-classifcation without alignment with a soft mask on SPT6. Particles containing SPT6 were combined (20.6%), reverted to original particles and 3D refined with a soft mask covering EC-DSIF-SPT6-U1 snRNP. This resulted in the final overall map of EC-DSIF-SPT6-U1 snRNP with 52,065 particles at 3.5 Å. To improve the local resolution of SPT6 and U1 snRNP, soft masks were applied individually onto SPT6Core and U1 snRNP. Following particle subtraction in RELION, particles were imported into CryoSPARC40 for local refinement followed by Global CTF refinement and local refinement, resulting in a 6.2 Å map for SPT6Core and a 7.3 Å map for U1 snRNP. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 29 Model building and refinement Initial models of Pol II (PDB: 7B0Y13) and DSIF (PDB: 9HVQ28) were rigid-body fitted into the overall map in Chimera41. Initial model of U1 snRNP (PDB: 7B0Y13) was rigid-body fitted into the focused refined map of U1 snRNP, while the model of SPT6 (PDB: 9HVQ28) was fitted into the focused refined map of SPT6. The Pol II model was manually adjusted in Coot42. The resulting complete model of EC-DSIF-SPT6-U1 snRNP was then real-space refined in the locally filtered and sharpened overall map using structure restraints of U1 snRNP, DSIF and SPT6 in PHENIX43. Figures were generated using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.) and Chimera X44. Data availability The cryo-EM reconstructions and final model were deposited with the EMDB under accession codes EMD-53087 (overall map), EMD-53088 (focused refined map of SPT6), EMD-53089 (focused refined map of U1 snRNP) and the PDB under accession code 9QEQ.

references 32. Gradia SD, Ishida JP, Tsai MS, Jeans C, Tainer JA, Fuss JO. MacroBac: New Technologies for Robust and Efficient Large -Scale Production of Recombinant Multiprotein Complexes. Methods Enzymol 592, 1-26 (2017). 33. Liu H, Naismith JH. An efficient one -step site-directed deletion, insertion, single and multiple-site plasmid mutagenesis protocol. BMC Biotechnol 8, 91 (2008). 34. Sakuma T, Nakade S, Sakane Y, Suzuki KT, Yamamoto T. MMEJ -assisted gene knock-in using TALENs and CRISPR -Cas9 with the PITCh systems. Nat Protoc 11, 118-133 (2016). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 30 35. Zumer K, et al. FACT maintains chromatin architecture and thereby stimulates RNA polymerase II pausing during transcription in vivo. Mol Cell 84, 2053 -2069 e2059 (2024). 36. Mastronarde DN. Automated electron microscope tomography using robust prediction of specimen movements. J Struct Biol 152, 36-51 (2005). 37. Tegunov D, Cramer P. Real -time cryo-electron microscopy data preprocessing with Warp. Nat Methods 16, 1146-1152 (2019). 38. Scheres SH. A Bayesian view on cryo -EM structure determination. J Mol Biol 415, 406-418 (2012). 39. Scheres SH. Processing of Structurally Heterogeneous Cryo -EM Data in RELION.

Enzymol 579, 125-157 (2016). 40. Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods 14, 290-296 (2017). 41. Pettersen EF, et al. UCSF Chimera--a visualization system for exploratory research and analysis. J Comput Chem 25, 1605-1612 (2004). 42. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66, 486-501 (2010). 43. Liebschner D, et al. Macromolecular structure determination using X -rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol 75, 861- 877 (2019). 44. Goddard TD, et al. UCSF ChimeraX: Meeting modern challenges in visualization and analysis. Protein Sci 27, 14-25 (2018). .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint 31 Acknowledgments We thank all members of the Zhang labs for discussion. We thank C. Dienemann and U. Steuerwald for support at the microscope, T. Schulz for the pig thymus, J. Schmitzova for initial trials of Pol II purification. We thank the LMB scientific computing for maintaining the computing cluster and K. Turton for maintaining the insect cell facility. P.C. is supported by the Max Planck Society. S.Z. is funded by the Medical Research Council, as part of United Kingdom Research and Innovation (also known as UK Research and Innovation) with the MRC file reference number MC_UP_1201/30. For open access, the MRC Laboratory of Molecular Biology has applied a CC BY public copyright license to any Author Accepted Manuscript version arising. Author contributions L.Z. purified proteins and performed biochemistry experiments. C.B. performed ITC experiments. S.A. assisted data processing. Y.G. purified proteins. K.Z. and K.M. generated the Pol II cell line. P.C. and S.Z. supervised the project. S.Z. designed and performed experiments, collected and analyzed cryo-EM data, and wrote the manuscript with input from all other authors. Competing interests Authors declare that they have no competing interests. Additional information Supplementary Information Correspondence and request of materials should be addressed to S.Z. under a material transfer agreement with the MRC Laboratory of Molecular Biology. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 21, 2025. ; https://doi.org/10.1101/2025.03.21.644610doi: bioRxiv preprint