Paf1 and Ctr9 subcomplex formation is essential for Paf1 complex assembly and functional regulation AbstractThe evolutionarily conserved multifunctional polymerase-associated factor 1 (Paf1) complex (Paf1C), which is composed of at least five subunits (Paf1, Leo1, Ctr9, Cdc73, and Rtf1), plays vital roles in gene regulation and has connections to development and human diseases. Here, we report two structures of each of the human and yeast Ctr9/Paf1 subcomplexes, which assemble into heterodimers with very similar conformations, revealing an interface between the tetratricopeptide repeat module in Ctr9 and Paf1. The structure of the Ctr9/Paf1 subcomplex may provide mechanistic explanations for disease-associated mutations in human PAF1 and CTR9. Our study reveals that the formation of the Ctr9/Paf1 heterodimer is required for the assembly of yeast Paf1C, and is essential for yeast viability. In addition, disruption of the interaction between Paf1 and Ctr9 greatly affects the level of histone H3 methylation in vivo. Collectively, our results shed light on Paf1C assembly and functional regulation. IntroductionThe highly conserved eukaryotic multifunctional polymerase-associated factor 1 (Paf1) complex (Paf1C) was originally identified in budding yeast over 2 decades ago1, and robust data suggested that it plays a vital role in transcriptional elongation and chromatin modifications2,3. Yeast Paf1C comprises subunits Paf1, Leo1, Ctr9, Cdc73, and Rtf14, while the human homolog PAF1C (human PAF1C and each subunit are written in all caps to distinguish yeast Paf1C and subunits, hereafter) contains an additional subunit, SKI8, which belongs to the SKI complex known for RNA quality control5. PAF1C is recruited to the transcription machinery by CTR96 or CDC737 binding to RNA polymerase II (pol II). In addition to its interactions with pol II, Paf1C has extensive genetic and physical links to yeast elongation factors Spt4/Spt5, Spt16/Pob3, and Dst1 (corresponding to human DSIF, FACT, and TFIIS/SII, respectively)4,6,8, allowing it to regulate transcription elongation procession.Paf1C or PAF1C is also involved in transcription-coupled histone modifications. Paf1C or PAF1C is required to promote H2B ubiquitination, as well as H3 methylation at K4, K36, and K79 in yeast, flies, and humans5,9,10,11,12,13. Paf1C or PAF1C participates in mRNA 3\'-end processing14,15,16,17,18, and nuclear export of transcripts19, thereby linking transcription and posttranscriptional regulations. Paf1C or PAF1C participates in such other cellular functions as cell cycle regulation20, stem cell pluripotency maintenance21, signal transduction22, DNA repair23,24, small-RNA-mediated gene silencing25 or activation26, and the regulation of promoter-proximal pausing of pol II27,28, or the release of paused pol II29,30,31, depending on context.PAF1C has also been implicated in tumorigenesis3. Thus, subunits of PAF1C could act as oncogenes or tumor suppressors in a context-dependent manner. For example, the overexpression of PAF1 promotes pancreatic cancer formation32, and regulates self-renewal and drug resistance of pancreatic and ovarian cancer stem cells33. In contrast, germline mutations in CTR9 predispose to Wilms tumor34, suggesting that CTR9 may act as a tumor suppressor. Furthermore, Wei Xu and coworkers discovered that CTR9 is a central regulator of estrogen signaling that drives ER伪+ breast tumorigenesis35. CDC73, encoded by HRPT2, is a tumor suppressor that is mutated in the germline of hyperparathyroidism鈥搄aw tumor syndrome (OMIM 145001)7,36. Interestingly, CDC73 is amplified in liver carcinoma and breast cancer37.The structures of Ras-like domain of Cdc7338,39 and the Plus3 domain of RTF140,41,42 provide the structural basis for Paf1 complex chromatin association. Recently, the crystal structure of the N-terminal domain of CDC73 has been resolved, which may provide the molecular mechanisms of hyperparathyroidism鈥搄aw tumor mutants43. The structure of histone modification domain (HMD) of Rtf1 was reported and the HMD was shown to stimulate H2B ubiquitylation through interaction with Rad644. However, there is no atomic structure information has been reported about CTR9, which contains multimotifs for protein鈥損rotein interaction (Fig.聽1a).Fig. 1Crystal structure of the human CTR9/PAF1 heterodimer. a Schematic representation of full-length CTR9 and PAF1. A LEO1-interacting region (blue) and a histone H3-interacting region (orange) are shown in PAF1. The predicted 19 TPR motifs (gray) were defined using TPRpred (https://toolkit.tuebingen.mpg.de/#/tools/tprpred). The protein fragments of the CTR9(1鈥?49)/PAF1(57鈥?16) complex used for structural determination are indicated by a two-way arrow and are colored cyan and magenta, respectively. b Ribbon diagram representation of the CTR9(1鈥?49) (cyan)/PAF1(57鈥?16) (magenta) complex viewed from the side. The N- and C-termini of the two proteins are labeled. Cylinder representation of the CTR9(1鈥?49)/PAF1(57鈥?16) complex structure viewed from the top (c) or the bottom (d). e Surface representation of the CTR9(1鈥?49)/PAF1(57鈥?16) complex with its orientation corresponding to the other side from that shown in (b)Full size imageTo date, structural information on yeast Paf1C or human PAF1C assembly is limited, though an architecture of yeast Paf1C with a local resolution of ~10鈥?0鈥壝?was obtained recently45. In a previous study, we determined the crystal structure of the human PAF1/LEO1 heterodimer and showed that CTR9 acts as a key scaffold protein during PAF1C assembly46. In this study, we determine the atomic structures of both the human PAF1/CTR9 and yeast Paf1/Ctr9 heterodimers, both with a resolution of ~2.53鈥壝? Our results clearly reveal that the heterodimer of PAF1/CTR9 or Paf1/Ctr9 is the core component and is important for human PAF1C or yeast Paf1C assembly, yeast viability, and Paf1C-mediated histone modifications. Our study suggests that the PAF1/CTR9 or Paf1/Ctr9 subcomplex-mediated holocomplex assembly and functional regulation may be a general mechanism for complexes from other eukaryotic species.ResultsThe interaction between human CTR9 and PAF1Structural and biochemical data reveal that LEO1 binds to human PAF1C through PAF1 and that the CTR9 subunit is the key scaffold protein in assembling PAF1C46. Protein sequence analysis showed that both human CTR9 (Fig.聽1a) and yeast Ctr9 (see below) possessed multiple tetratricopeptide repeat (TPR) motifs. We confirmed that the human CTR9, PAF1, and LEO1 (herein named CTR9/PAF1/LEO1, 鈥?鈥?denotes protein complexes with separate chains, similar nomenclature hereafter) subunits interacted by showing that three thioredoxin (Trx)-tagged fusion proteins [aa 1鈥?84, Trx-CTR9(1鈥?84), Trx-PAF1, and Trx-LEO1] eluted from the size-exclusion column as a tripartite complex (Supplementary Fig.聽1a). Next, we sought to test the heteromeric complex formation between PAF1 and CTR9(1鈥?84). We found that a conserved N-terminal fragment of PAF1 (aa 57鈥?66, PAF1(57鈥?66)) formed a stable dimeric complex with CTR9(1鈥?84) (Supplementary Fig.聽1b). Using a similar truncation-based approach, we mapped the minimal PAF1-binding region, including TPR1鈥? of CTR9, to a 249-residue fragment (aa 1鈥?49, CTR9(1鈥?49)) and the minimal CTR9-binding region of PAF1 to a 60-residue fragment (aa 57鈥?16, PAF1(57鈥?16)) (Fig.聽1a and Supplementary Fig.聽1c鈥揺). It was noted that CTR9(1鈥?49) and PAF1(57鈥?16) assembled into a complex since their co-elution in the analytical size-exclusion column (red line and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) inset in Supplementary Fig.聽1e). Analytical ultracentrifugation further confirmed that the CTR9(1鈥?49)/PAF1(57鈥?16) complex formed a stoichiometry of a 1:1 heterodimer with a molecular mass of ~34鈥塳Da (red line in Supplementary Fig.聽1f). More, to further validate the role of PAF1(57鈥?16) in CTR9/PAF1 complex formation, the N-terminal 116 amino acids of PAF1 were deleted in mutant [referred to as PAF1(N116螖)]. Notably, green fluorescent protein (GFP)-tagged PAF1(N116螖) [GFP-PAF1(N116螖)] mutant did not co-immunoprecipitate with Myc-tagged CTR9 (Myc-CTR9) thoroughly (lane 3 in Supplementary Fig.聽1g), whereas GFP-PAF1 specifically formed a complex with Myc-CTR9 (lane 2 in Supplementary Fig.聽1g) in a co-immunoprecipitation (co-IP) assay, indicating that the N-terminal fragment including amino acids 57鈥?16 of PAF1 is essential for the interaction between full-length CTR9 and PAF1. Given these results, we conclude that the CTR9/PAF1 subcomplex forms a heterodimer through interaction between fragments CTR9(1鈥?49) and PAF1(57鈥?16).Crystal structure of the CTR9 and PAF1 heterodimerTo understand how CTR9 and PAF1 bind to each other, we attempted to determine the crystal structure of the CTR9(1鈥?49)/PAF1(57鈥?16) heterodimer using two separate chains; these experiments were unsuccessful. However, we succeeded in obtaining crystals of the single polypeptide created by the fusion of CTR9(1鈥?49) to the N-terminus of PAF1(57鈥?16) with a tobacco etch virus (TEV)-cleavable segment. This technique has been used previously to determine the PAF1/LEO1 heterodimer structure46. The purified single-chain fusion protein of CTR9(1鈥?49) and PAF1(57鈥?16) (herein named CTR9(1鈥?49)鈥揚AF1(57鈥?16), 鈥?鈥?denotes proteins in a single-chain fusion, similar nomenclature hereafter) eluted as a single peak from an analytical size-exclusion column (black line in Supplementary Fig.聽1e) and assembled into a heterodimer with a molecular mass of ~35鈥塳Da from the sedimentation velocity (SV) experiment (black line in Supplementary Fig.聽1f). The crystal structure of CTR9(1鈥?49)鈥揚AF1(57鈥?16) was determined at a resolution of 2.53鈥壝?and with one molecule per asymmetric unit (Supplementary Table聽1). In the final model, all residues were clearly visible, except residues 1鈥? and 246鈥?49 of CTR9(1鈥?49) and residues 57鈥?6 and 110鈥?16 of PAF1(57鈥?16), presumably owing to conformation flexibility (Fig.聽1b鈥揺, Fig.聽2e, and Supplementary Figs.聽2 and 3a). Interestingly, we noted that the entire artificial TEV-cleavable linker was missing in the final model (Supplementary Fig.聽3b). This was seemingly caused by the existence of multiple conformers of the linker in the current structure because SDS-PAGE analysis of dissolved CTR9(1鈥?49)鈥揚AF1(57鈥?16) crystals demonstrated that the molecules were intact (Supplementary Fig.聽3c). These data suggested that the artificial linker did not alter the global structure of the CTR9(1鈥?49)鈥揚AF1(57鈥?16) and may stabilize the N-terminus of PAF1(57鈥?16) to help complex proteins crystallization.Fig. 2The interaction interface of the human CTR9/PAF1 heterodimer. a鈥?b>e The CTR9(1鈥?49)/PAF1(57鈥?16) interface is divided into three regions corresponding to the CTR9 TPR1鈥? concave channel/N-terminal loop of PAF1 (b, e), the side of the TPR1鈥?/C-terminal loop of PAF1 (c, e), and the convex surface of the TPR4鈥?/伪D helix of PAF1 (d, e). The interaction details between CTR9 and PAF1 in the three regions are shown in (b) to (d). Charge鈥揷harge or hydrogen-bonding and hydrophobic interactions are shown as gray dotted lines and spoked arcs, respectively. e Structural-based sequence alignment of the CTR9-binding fragments of PAF1 in various species. In this alignment, the secondary structures of human PAF1 and yeast Paf1 are shown at the top and bottom, respectively, according to the crystal structures of CTR9(1鈥?49)鈥揚AF1(57鈥?16) and Ctr9(1鈥?13)鈥揚af1(34鈥?03), and conserved residues are shaded in red. The highly conserved residues, which are mutated in PAF1(57鈥?16)(4S), PAF1(57鈥?16)(D104K), or PAF1(57鈥?16)(5M) and the Paf1(4S), Paf1(L83S) or Paf1(D95K) construct, are indicated with magenta and green asterisks, respectively. The disease-associated amino acid substitutions I87M or E109K (category 1 shown in Fig.聽3a) and D85N or D95Y (category 2 shown in Fig.聽3a) are indicated with red and orange spheres, respectively. Three regions of CTR9-binding in PAF1 are indicated with dotted boxes. Species abbreviations: H.s, Homo sapiens; M.m, Mus musculus; D.r, Danio rerio; D.m, Drosophila melanogaster; C.e, Caenorhabditis elegans; S.c, Saccharomyces cerevisiae. f Co-IP experiments testing the interaction between CTR9(1鈥?49) and PAF1(57鈥?16) wild-type (WT) or mutants. The PAF1(57鈥?16)(4S) mutant contains four amino acid substitutions L80S, V82S, I84S, and L86S. The PAF1(57鈥?16)(5M) mutant contains five amino acid substitutions L80S, V82S, I84S, L86S, and D104K. Myc was tagged to the N-terminal of CTR9(1鈥?49) and GFP was tagged to the C-terminal of PAF1(57鈥?16) WT or mutant. Extracts were prepared from HEK293T cells transfected with various combinations of plasmids, as indicated. The bottom panel shows 3% of the Myc fusion proteins for each IP. Uncropped blots are shown in Supplementary Fig.聽8Full size imageThe structure of CTR9(1鈥?49) was composed of 12 伪-helices and two 尾-strands at the N-terminus (Fig.聽1b鈥揹 and supplementary Fig.聽2). The helices 伪1A to 伪5B contained five sequential helix-turn-helix repeats that exhibited structural similarity to TPR (Fig.聽1b鈥揹 and Supplementary Fig.聽4a). All five of these TPR motifs, named TPR1 (伪1A and 伪1B), TPR2 (伪2A and 伪2B), TPR3 (伪3A and 伪3B), TPR4 (伪4A and 伪4B), and TPR5 (伪5A and 伪5B), were conserved with the canonical TPR motif, except TPR2, which had both a longer A helix and B helix and was distinct from the canonical TPR motif containing 34 amino acids and other noncanonical TPR motifs (Supplementary Fig.聽4b and refs. 47 and 48). Like multiple other TPR motifs49,50,51,52,53, the five TPR units of CTR9(1鈥?49) were arranged in parallel with adjacent 伪-helices to each other to create a right-handed superhelix (Fig.聽1b鈥揹). It was also noted that there was an N-terminal cap composed of the antiparallel 尾-sheet (尾1 and 尾2) and the 伪0 helix and a C-terminal cap composed of the 伪6A helix, which interacted with TPR1 and TPR5, respectively (Fig.聽1b鈥揹). The PAF1(57鈥?16) structure, which was composed of a long loop flanked by two 伪 helices (伪A and 伪D) in the N- and C-terminus, respectively, adopted a hook-shaped conformation occupying the entire concave channel formed by subdomains of TPR1鈥? and TPR4鈥? and the convex surface formed by TPR4鈥? (Fig.聽1b鈥揹).CTR9 and PAF1 interactions in the heterodimerThe CTR9/PAF1 interface could be divided into three regions according to the heterodimer structure (Fig.聽2a, e): region I: the N-terminal half-loop of PAF1(57鈥?16) and the concave channel of TPR1鈥? (Fig.聽2b); region II: the C-terminal half-loop of PAF1(57鈥?16) and the side of TPR1鈥? (Fig.聽2c); and Region III: the 伪D helix at the C-terminus of PAF1(57鈥?16) and the convex surface of TPR4鈥? (Fig.聽2d). Region I was mainly maintained through four conserved hydrophobic amino acids (L80, V82, I84, and L86) in PAF1 and corresponding amino acids in CTR9 by hydrophobic interaction (Fig聽2b, e, and Supplementary Fig.聽5a). Region II was mainly mediated by two types of interactions between amino acid pairs from PAF1 and CTR9, respectively: hydrogen bond in D90/K53 pair and hydrophobic interactions (such as P89, T91, T92, I94, and V98 in PAF1 and corresponding amino acids in CTR9) (Fig.聽2c and Supplementary Fig.聽5b). The binding of Region III was largely mediated through hydrogen bonds by several pairs of amino acids between PAF1 and CTR9: D104/R201, D104/R223, and K106/K216, which formed an extensive network of hydrogen bonds (Fig.聽2d and Supplementary Fig.聽5c).We performed a series of mutagenesis studies to validate the interactions observed in the structure of the CTR9(1鈥?49)鈥揚AF1(57鈥?16) complex. To this end, four conserved hydrophobic residues from region I (L80, V82, I84, and L86) were substituted with serine in PAF1(57鈥?16) [referred to as PAF1(57鈥?16)(4S)], residue D104 from region III, which is conserved from yeast to human, was mutated to lysine in PAF1(57鈥?16) [referred to as PAF1(57鈥?16)(D104K)], and all five amino acids substitutions L80S, V82S, I84S, L86S, and D104K were included in PAF1(57鈥?16) [referred to as PAF1(57鈥?16)(5鈥塎)]. Consistent with the structure-based prediction, the purified PAF1(57鈥?16)(5鈥塎) could not form a stable complex with CTR9(1鈥?49) in the analytical size-exclusion column (dashed line in Supplementary Fig.聽6a and SDS-PAGE in Supplementary Fig.聽6a2). Interestingly, it was noted that CTR9(1鈥?49) lacking of interaction with the PAF1(57鈥?16)(5鈥塎) mutant or without PAF1(57鈥?16) is aggregated in size-exclusion column (dotted line and blue line comparing to black line in Supplementary Fig.聽6a and SDS-PAGEs in Supplementary Fig.聽6a2 and a3 comparing to a1), indicating that CTR9(1鈥?49)/PAF1(57鈥?16) heterodimer formation is important for maintaining the conformation of CTR9(1鈥?49). More, consistent with this result, the C-terminal GFP-tagged mutants PAF1(57鈥?16)(5M), PAF1(57鈥?16)(D104K), and PAF1(57鈥?16)(4S) [PAF1(57鈥?16)(5M)-GFP, PAF1(57鈥?16)(D104K)-GFP, and PAF1(57鈥?16)(4S)-GFP] did not co-immunoprecipitate with Myc-tagged CTR9(1鈥?49) (Myc-CTR9(1鈥?49)) (lanes 3鈥? in Fig.聽2f), whereas PAF1(57鈥?16)-GFP WT specifically formed a complex with Myc-CTR9(1鈥?49) (lane 2 in Fig.聽2f) in a co-IP assay. Additionally, the structure of CTR9(1鈥?49)鈥揚AF1(57鈥?16) indicated that the 伪A of PAF1(57鈥?16) may be not important for CTR9(1鈥?49)/PAF1(57鈥?16) heterodimer formation (Fig.聽2). To further validate this observation, a truncated fragment of PAF1 (aa 75鈥?16, PAF1(75鈥?16)) was designed to delete the N-terminal 伪A of PAF1(57鈥?16). As expected, size-exclusion chromatography and SDS-PAGE analysis showed that PAF1(75鈥?16) formed a complex with CTR9(1鈥?49) (cyan line and SDS-PAGE inset in Supplementary Fig.聽1h). However, a temperature-dependent denaturation assay demonstrated that the CTR9(1鈥?49)/PAF1(75鈥?16) complex is less stable than the CTR9(1鈥?49)/PAF1(57鈥?16) complex (cyan line comparing to red line and insert in Supplementary Fig.聽1i), indicating that the 伪A of PAF1(57鈥?16) may contribute to the stability of the CTR9(1鈥?49)/PAF1(57鈥?16) complex. Collectively, these results confirm the interaction mode revealed by the structure of CTR9(1鈥?49)鈥揚AF1(57鈥?16).Disease-associated mutations affect CTR9 binding to PAF1The deletion of the CTR9 gene locus with PAF1 locus amplification may synergistically lead to initiation and progression of pancreatic cancer37. A total of 38 and 17 missense mutations were located in 38 residues of CTR9(1鈥?49) and 12 residues of PAF1(57鈥?16) (Supplementary Tables聽2 and 3), respectively, based on the data extracted from the COSMIC database (http://cancer.sanger.ac.uk/cosmic). We then mapped the 55 mutations to the modeled structure and divided these mutations into three categories (named as interface, folding, and others, respectively, only mutations used in this study were shown in Fig.聽3a and the full list of mutations was shown in Supplementary Table聽2 or 3). The most interesting category (interface), which was directly relevant to this study, contained five mutations (P164S, R201C, and A220T of CTR9(1鈥?49) and I87M and E109K of PAF1(57鈥?16)) located at the interface of the CTR9(1鈥?49)鈥揚AF1(57鈥?16) heterodimer. Category 2 (folding) included mutations that were likely to affect the folding of each individual subunit. The third category (others) of mutations could not be well explained by the current structural model.Fig. 3Disease-associated mutations affect the interaction between CTR9 and PAF1. a Disease-associated mutations in the CTR9(1鈥?29)/PAF1(57鈥?16) complex. For clarify, only five missense-mutation sites in category 1 (interface), two sites in category 2 (folding) of PAF1, and one site (R98W) in category 3 (others) of CTR9 are highlighted with spheres and colored in red, orange, and green, respectively. The full lists of disease-associated mutations in CTR9 and PAF1 are summarized in Supplementary Tables聽2 and 3, respectively. b The interaction sites between CTR9(1鈥?49) and PAF1(57鈥?16) containing various disease-associated mutations were evaluated using a co-IP strategy. Myc was tagged to the N-terminal of CTR9(1鈥?49) WT or mutant and GFP was tagged to the C-terminal of PAF1(57鈥?16) WT or mutant. Extracts were prepared from HEK293T cells transfected with various combinations of plasmids, as indicated. The bottom panel shows 3% of the Myc fusion proteins for each IP. Uncropped blots are shown in Supplementary Fig.聽8Full size imageGiven the extremely high conservation of these five amino acids located at the interface of the CTR9(1鈥?49)鈥揚AF1(57鈥?16) heterodimer from zebrafish to human (Fig.聽2e and Supplementary Fig.聽2), we next sought to test the effects of these disease category 1 (interface) mutations on the formation of the CTR9(1鈥?49)/PAF1(57鈥?16) heterodimer. To this end, using the co-expression strategy, we noted that the purified mutant PAF1(57鈥?16)(I87M) or PAF1(57鈥?16)(E109K) did not form a stable complex with CTR9(1鈥?49) (magenta lines in Supplementary Fig.聽6b and SDS-PAGEs in Supplementary Fig.聽6b1 and b2), nor did any purified CTR9(1鈥?49) mutant (P164S, R201C, or A220T) form a stable complex with PAF1(57鈥?16) (dashed cyan line or orange lines in Supplementary Fig.聽6c and SDS-PAGEs in Supplementary Fig.聽6c2-c4), in the analytical size-exclusion column. It was also noted that CTR9(1鈥?49) WT or mutants are aggregated in size-exclusion column in the absence of PAF1(57鈥?16) binding (Supplementary Fig.聽6b and 6c). More, in the co-IP assays, it was clear that the I87M and E109K mutations of PAF1 and the P164S, R201C, and A220T mutations of CTR9 totally abolished the interaction between CTR9(1鈥?49) and PAF1(57鈥?16) (Fig.聽3b). As a negative control, the category 3 (others) R98W mutation of CTR9 did not affect the interaction between CTR9(1鈥?49) and PAF1(57鈥?16), since the purified CTR9(1鈥?49)(R98W) mutant formed a stable complex with PAF1(57鈥?16) (cyan line in Supplementary Fig.聽6c and SDS-PAGE in Supplementary Fig.聽6c1) and the Myc-CTR9(1鈥?49)(R98W) mutant specifically formed a complex with PAF1(57鈥?16)-GFP (lane 7 in Fig.聽3b).Overall structure of the yeast Ctr9 and Paf1 heterodimerPAF1C is conserved throughout eukaryotes in its subunit compositions and functional importance2,3,37, indicating that the structural features of each subunit and the assembly modes of holo- and/or sub-PAF1C are highly conserved among all eukaryotes. To evaluate this speculation, we investigated the structural features of the Ctr9/Paf1 heteromeric complex from Saccharomyces cerevisiae because PAF1C was initially identified in this species1. Accordingly, we analyzed the secondary structure predictions of yeast Ctr9 and Paf1. These predictions plus sequence alignment with the interaction regions of the human CTR9/PAF1 heterodimer (Fig.聽2e and Supplementary Fig.聽2), helped us identify the cognate interaction regions of the Ctr9/Paf1 complex through a 313-residue fragment of Ctr9 (aa 1-313, Ctr9(1鈥?13)), including TPR1-6, and a 70-residue fragment of Paf1 (aa 34鈥?03, Paf1(34鈥?03)) (Fig.聽4a). It was noted that Ctr9(1鈥?13) and Paf1(34鈥?03) assembled into a complex in the analytical size-exclusion column (red line in Supplementary Fig.聽7a and SDS-PAGE in Supplementary Fig.聽7a1). Next, we used the same approach by creating a single polypeptide through fusing Ctr9(1鈥?13) to the N-terminus of Paf1(34鈥?03) with a TEV-cleavable segment. The purified single-chain fusion protein of Ctr9(1鈥?13) and Paf1(34鈥?03) (referred to as Ctr9(1鈥?13)-Paf1(34鈥?03) hereafter) eluted as a single peak from an analytical size-exclusion column (black line in Supplementary Fig.聽7a and SDS-PAGE in Supplementary Fig.聽7a2), and subsequently, the crystals of Ctr9(1鈥?13)鈥揚af1(34鈥?03) were successfully obtained.Fig. 4Overall structure of the yeast Ctr9/Paf1 heterodimer. a Schematic representation of full-length Ctr9 and Paf1. The 20 predicted TPR motifs (gray) were defined using TPRpred. The protein fragments of the Ctr9(1鈥?13)/Paf1(34鈥?03) complex used for structural determination are indicated by a two-way arrow and are colored orange and green, respectively. b Cylinder representation of the Ctr9(1鈥?13) (orange)/Paf1(34鈥?03) (green) complex viewed from the side. The N- and C-termini of the two proteins are labeled. The Ctr9(1鈥?13)/Paf1(34鈥?03) complex structure viewed from the top (c) or the bottom (d). e鈥?b>g The Ctr9(1鈥?13)/Paf1(34鈥?03) interface is divided into three regions corresponding to the Ctr9 TPR1鈥? concave channel/N-terminal loop of Paf1 (e and Fig.聽2e), the side of the TPR1鈥?/C-terminal loop of Paf1 (f and Fig.聽2e), and the convex surface of the TPR4鈥?/伪D\' helix of Paf1 (g and Fig.聽2e). The interaction details between Ctr9 and Paf1 in the three regions are shown in e鈥?b>g. Charge鈥揷harge or hydrogen-bonding and hydrophobic interactions are shown as gray dotted lines and spoked arcs, respectively. h Co-IP experiments testing the interactions between Ctr9 and Paf1-WT or mutants. The Paf1(4S) mutant contains four amino acid substitutions L62S, M64S, V66S, and L68S. Extracts were prepared from HEK293T cells transfected with various combinations of plasmids, as indicated. Myc and GFP were tagged to the Paf1 and Ctr9, respectively. The bottom panel shows 3% of the Myc fusion proteins for each IP. Uncropped blots are shown in Supplementary Fig.聽8Full size imageThe crystal structure of Ctr9(1鈥?13)鈥揚af1(34鈥?03) was determined at a resolution of 2.53鈥壝?with one molecule in the asymmetric unit (Supplementary Table聽1). The entire artificial TEV-cleavable linker was also missing in the final model (Supplementary Fig.聽3d and f), though SDS-PAGE analysis of dissolved native and Se-Met Ctr9(1鈥?13)鈥揚af1(34鈥?03) crystals showed that these molecules were intact (Supplementary Fig.聽3e). These results suggest that the artificial linker also did not alter the global structure of the Ctr9(1鈥?13)鈥揚af1(34鈥?03). Notably, sequence alignment of Ctr9(1鈥?13) and CTR9(1鈥?49) indicated that the structure of yeast Ctr9(1鈥?13) had two additional 伪-helices (伪6B\' and 伪7A\') compared to human CTR9(1鈥?49) (Supplementary Fig.聽2), suggesting that these two 伪 helices are not necessary for Ctr9 binding to Paf1. To test this hypothesis, a truncated fragment of Ctr9 (aa 1鈥?70, Ctr9(1鈥?70)) was designed to test its binding to Paf1(34鈥?03). As expected, size-exclusion chromatography and SDS-PAGE analysis showed that Ctr9(1鈥?70) formed a heteromeric complex with Paf1(34鈥?03) (blue line in Supplementary Fig.聽7b and SDS-PAGE in Supplementary Fig.聽7b1). Analytical ultracentrifugation further confirmed that the Ctr9(1鈥?70)/Paf1(34鈥?03) complex formed a heterodimer with a stoichiometry of a 1:1 and with a molecular mass of ~36鈥塳Da (Supplementary Fig.聽7c).Superimposition of the structures of the yeast Ctr9(1鈥?13)鈥揚af1(34鈥?03) and human CTR9(1鈥?49)鈥揚AF1(57鈥?16) heterodimers resulted in a root-mean-square deviation of 2.63鈥壝?for 198 equivalent C伪 atoms (Supplementary Fig.聽4c), indicating that human CTR9/PAF1 and yeast Ctr9/Paf1 subcomplexes share similar heterodimer structures. Notably, the sequence identity was low between the yeast Ctr9 or Paf1 and human CTR9 or PAF1 subunits (Fig.聽2e and Supplementary Fig.聽2); thus, the residues at each heterodimer interface were divergent (Figs.聽2b鈥揺 and 4e鈥揼). Similar to human CTR9/PAF1 subcomplex interactions, the Ctr9(1鈥?13)/Paf1(34鈥?03) interface could be divided into three regions (Figs.聽4e鈥揼 and 2e). In detail, region I was also mainly maintained by hydrophobic interaction through four conserved hydrophobic amino acids in Paf1 and corresponding amino acids in Ctr9, and by hydrogen bonds between D61/Q150 and K191, M64/K191, and D67/N104 (Fig.聽4e and Supplementary Fig.聽5d). Substitution of these four amino acids in Paf1 with serines (L62S, M64S, V66S, and L68S, referred to 4S) led to the complete disruption of Paf1/Ctr9 subcomplex formation (lane 3 in Fig.聽4h). Region II of Paf1 was not conserved between yeast and other species (Fig.聽2e). In this region, Paf1 had a longer loop than other species (e.g., worm to human), including a shorter 伪-helix (伪C\') that bound the N-terminal tail of Ctr9 (Fig.聽4f and Supplementary Fig.聽5e), which is different from human CTR9/PAF1 complex. We noted that yeast Ctr9 had a longer N-terminal tail than other species (e.g., human CTR9) (Fig.聽5a). In order to test the role of this longer N-terminal tail of Ctr9 in Ctr9/Paf1 complex formation, the most N-terminal 16 amino acids of Ctr9 were deleted in mutant [Ctr9(N16螖)]. Notably, the [Ctr9(N16螖)] mutant did not form complex with Paf1 thoroughly (Fig.聽5b), indicating that this longer N-terminal tail of Ctr9 is essential to maintain the interaction between Ctr9 and Paf1. We also noted that residue L83 of Paf1 is only present in yeast (Fig.聽2e), though L83 is involved in binding to Ctr9 (Fig.聽4f). Substitution of L83 with serine (L83S) in Paf1 was not enough to disrupt the formation of the Ctr9/Paf1 subcomplex (lane 4 in Fig.聽4h), indicating that residue L83 of Paf1 is not important for its interaction with Ctr9. For region III, D95 was the core amino acid in the C-terminus 伪D\' of Paf1, which bound with R221, R243, and Y10 of Ctr9, mainly mediated by extensive hydrogen bonds (Fig.聽4g and Supplementary Fig.聽5f). D95K mutation in Paf1 also abolished the interaction between Ctr9 and Paf1 completely (lane 5 in Fig.聽4h). Similar to human CTR9/PAF1 structure, it was also noted that the 伪A\' of Paf1(34鈥?03) is not involved in Ctr9(1鈥?13)/Paf1(34鈥?03) heterodimer formation (Fig.聽4b鈥揹). Accordingly, a truncated fragment of Paf1 (aa 59鈥?03, Paf1(59鈥?03)) was designed to delete the 伪A\' of Paf1(34鈥?03). Size-exclusion chromatography and SDS-PAGE analysis showed that Paf1(59鈥?03) formed a complex with Ctr9(1鈥?13) (cyan line in Supplementary Fig.聽7b and SDS-PAGE in Supplementary Fig.聽7b2), and a temperature-dependent denaturation assay demonstrated that the Ctr9(1鈥?13)/Paf1(59鈥?03) complex is less stable than the Ctr9(1鈥?13)/Paf1(34鈥?03) complex (cyan line comparing to red line and inset in Supplementary Fig.聽7d), indicating that the 伪A\' of Paf1(34鈥?03) may contribute to the stability of the Ctr9(1鈥?13)/Paf1(34鈥?03) complex. These structural analyses and mutagenesis-based data suggest that the specific recognition between Ctr9 and Paf1 results from a combination of numerous hydrophobic, charge鈥揷harge, and hydrogen-bonding interactions along the Ctr9 TPRs and the corresponding residues of Paf1.Fig. 5The longer N-terminal tail of Ctr9 is essential for its binding to Paf1. a Structural-based sequence alignment of the N-terminal fragments of Ctr9 in various species. In this alignment, the secondary structures of human CTR9(1鈥?8) and yeast Ctr9(1鈥?3) are shown at the top and bottom, respectively, and conserved residues are shaded in red. The amino acids 1鈥?6 of yeast Ctr9, which were deleted in the GFP-Ctr9(N16螖) construct [used in b], are marked with a dotted blue box. The amino acids Y10, P11, M13, E14, and W15 of Ctr9 involved in its binding to Paf1 are marked by orange stars. b Co-IP experiments testing the interactions between Ctr9 WT or Ctr9(N16螖) mutant and Paf1. Extracts were prepared from HEK293T cells transfected with various combinations of plasmids, as indicated. The bottom panel shows 3% of the Myc fusion proteins for each IP. Uncropped blots are shown in Supplementary Fig.聽8Full size imageCtr9 and Paf1 subcomplex is essential for Paf1C assemblyPhenotypic analysis revealed that deletion either Paf1 or Ctr9 can influence yeast viability more severely than the other subunits of yeast Paf1C54,55, and CTR9 is considered a key scaffold protein in assembling human PAF1C46. These findings, as well as our results demonstrating the conserved heterodimeric assembly of both human CTR9 and PAF1 and yeast Ctr9 and Paf1, along with the finding that CTR9 and PAF1 are mutated in diseases, indicate that CTR9/PAF1 or Ctr9/Paf1 heterodimer formation may be important for complex assembly and function. To evaluate this hypothesis, we took yeast Paf1C as an example. To investigate the role of Ctr9/Paf1 subcomplex formation in the assembly of Paf1C, we used a co-IP strategy. In the WT Paf1C, both GFP-Ctr9 and GFP-Paf1 specifically co-immunoprecipitated with other Paf1C subunits (Fig.聽6a, lane 2, and Fig.聽6b, lane 2). In the mutant Paf1C, which contained mutants Myc-Paf1(4S) and Myc-Paf1(D95K), GFP-Ctr9 did not co-immunoprecipitate other Paf1C subunits (or amounts of some subunits to lesser degrees) (lanes 3 and 5 in Fig.聽6a). In another experiment, significantly decreased Myc-Ctr9, 13脳Myc-Rtf1, and Myc-Cdc73 and consistent amounts of Myc-Leo1 coprecipitated with both mutants GFP-Paf1(4S) and GFP-Paf1(D95K) (lanes 3 and 5 in Fig.聽6b). Notably, the mutant Paf1(L83S) (which was capable of binding to Ctr9), formed complexes with other Paf1C subunits (lanes 4 in Fig.聽6a, b). The asterisk in Fig.聽6b indicate the degradation of Myc-Ctr9, according to the data shown in Supplementary Fig聽7e. Collectively, these data indicate that Ctr9/Paf1 is the core component and this subcomplex formation is essential for assembling Paf1C. Together, these results, as well as the an architecture of Paf1C shown in prior study45, led to the establishment of the interaction network for the assembly of yeast Paf1C (Fig.聽6c).Fig. 6Ctr9/Paf1 subcomplex formation is essential for yeast Paf1C assembly. a, b Co-IP experiments of Paf1C formation by Ctr9 (a) and Paf1 (b). Extracts were prepared from HEK293T cells transfected with various combinations of plasmids, as indicated, immunoprecipitated with agarose-conjugated anti-GFP and subsequently immunoblotted with anti-Myc or anti-GFP, as indicated. The top panel shows the IP results. The middle panel represents the IP of GFP and GFP fusion proteins (GFP-Ctr9 or GFP-Paf1). The bottom panel shows 3% of the Myc fusion proteins for each IP. The asterisk indicate the degradation of Myc-Ctr9. c Model of yeast Paf1C assembly. The Ctr9/Paf1 heterodimer is the core component for Paf1C assembly. The bold line represents the interaction in the crystal structure of the Ctr9(1鈥?13)鈥揚af1(34鈥?03) heterodimer or the interaction between Paf1 and Leo1 obtained from the IP results and studies45,46. The fine lines represent the interaction obtained from the IP results. The dotted lines represent interactions that need to be further studied. Uncropped blots are shown in Supplementary Fig.聽8Full size imageCtr9 and Paf1 subcomplex is important for yeast viabilityThe described structural and biochemical results thus far clearly demonstrate that the conserved Ctr9/Paf1 subcomplex formation is important for the assembly of yeast Paf1C. Next, we used yeast as a host to investigate the role of the Ctr9/Paf1 subcomplex in vivo. To this end, a paf1螖 strain was generated by replacing the PAF1 gene with S. cerevisiae URA3 (Supplementary Table聽4). As expected, yeast cells with paf1螖 exhibited pleiotropic phenotypic traits, including slow growth, increased sensitivity to temperature, salt, and hydroxyurea (HU) (Fig.聽7a鈥揹), consistent with previous studies54,55,56. To further explore the role of the Ctr9/Paf1 subcomplex in vivo, we evaluated the importance of Ctr9/Paf1 subcomplex formation-mediated Paf1C assembly using a complementation assay. To evaluate their ability to complement paf1螖 phenotypes, the plasmids pP1K-PAF1(405), pP1K-PAF1(4S), pP1K-PAF1(L83S), and pP1K-PAF1(D95K) and the corresponding empty vectors were linearized and integrated into the paf1螖 strain. Interestingly, expression of both the WT PAF1 (PAF1-WT) and the mutant PAF1(L83S) (which was capable of assembling Paf1C), but not the mutant PAF1(4S) or PAF1(D95K) (which were incapable of assembling Paf1C), fully rescued the paf1螖 phenotype (Fig.聽7a鈥揹). Collectively, these results indicate that Ctr9/Paf1 subcomplex formation-mediated Paf1C assembly is crucial for yeast growth.Fig. 7Ctr9/Paf1 subcomplex formation is essential for the integrity of a functional yeast Paf1C. a鈥?b>d Ctr9/Paf1 subcomplex-mediated Paf1C assembly is essential for yeast cell viability. Strains of the indicated genotype were grown to late log phase/stationary phase (overnight) and plated in serial dilutions on a YPD plate. The plates are shown after 40鈥塰 of incubation at 30鈥壜癈 (a), at 37鈥壜癈 (b), on YPD containing 1鈥塎 NaCl (c), and on YPD containing 0.1鈥塎 hydroxyurea (HU) (d). e Representative immunoblot with specific H3K4me1/2/3 antibodies to detect methylation strength (top panels), or with H3 antibody to demonstrate equal loading (bottom panel). Bar graphs of H3K4me2 (f) and H3K4me3 levels (g) from various yeast strains, as indicated. All statistic data in this figure represents the results from three independent batches of experiments. Error bars represent s.e.m. (n鈥?鈥?). *P鈥?lt;鈥?.05, **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001. Uncropped blots are shown in Supplementary Fig.聽8Full size imageKnockout of Paf1 (paf1螖) in yeast cells leads to a strong, global decrease in synthesis of mRNA transcripts45, which might be required for normal growth. Our results showed that Ctr9/Paf1 subcomplex formation-mediated Paf1C assembly was important for the growth of yeast (Fig.聽7a鈥揹). These findings, together with the fact that Paf1C is required to promote H3 methylation at K4, K36, and K79 in yeast9,10,11, led us to consider whether the paf1螖 strain displayed pleiotropic phenotypes, including slow growth and increased sensitivity to temperature, salt, and HU because of a changed level of histone modification (e.g., H3K4 methylation). To test this hypothesis, we examined the level of histone methylation by analyzing the strength of the H3K4 methylation signal in whole yeast lysate. As expected, paf1螖 strains displayed drastically reduced levels of both the dimethylated and trimethylated H3K4 (referred to as H3K4me2 and H3K4me3, respectively), but not the monomethylated H3K4 (H3K4me1), compared to the wild-type strain (lane 2 compared with lane 1 in Fig.聽7e and the statistical data shown in Fig.聽7f, g). These data are consistent with previous studies9,10,11,57. The expression of both PAF1-WT and the mutant PAF1(L83S) genes, but not the mutant gene PAF1(4S) or PAF1(D95K) (both were incapable of Paf1C assembly), restored both the H3K4me2 and H3K4me3 levels against a paf1螖 background (lanes 3 and 5 compared with lanes 4 and 6 in Fig.聽7e and the statistical data shown in Fig.聽7f, g). Together, these data indicate that Ctr9/Paf1 subcomplex formation is indispensable for Paf1C-mediated histone modification(s).DiscussionThe crystal structures of human CTR9/PAF1 and yeast Ctr9/Paf1 subcomplexes presented in this study provide structural insights into the roles of Ctr9/Paf1 subcomplex formation-mediated Paf1C assembly and activity. To our knowledge, this is the paper to resolve crystal structures of the scaffold protein Ctr9 (albeit in part) at the atomic level. Though the two structures determined in this study covered only 21% and 29% of the residues of full-length human CTR9 and yeast Ctr9, respectively, we believe that such structural information will help us to probe into the CTR9-incorporated PAF1C architecture. Our analysis revealed that both the hook-shaped (hookfold) PAF1(57鈥?16) and Paf1(34鈥?03) occupied the concave channel and convex surface of the right-handed superhelix-folded CTR9(1鈥?49) and Ctr9(1鈥?13), respectively (Figs.聽1, 2, and 4), with very similar conformations (Supplementary Fig.聽4c), although the sequence identity was low between human PAF1 and yeast Paf1 (Fig.聽2e) or CTR9 and Ctr9 (Supplementary Fig.聽2). Therefore, it can be assumed that the structural and biochemical features of Paf1/Ctr9 subcomplex described herein are shared by the orthologs of Paf1 and Ctr9 in other eukaryotic species.Our previous study indicated that LEO1 binds to human PAF1C through PAF1 and that the CTR9 subunit is the key scaffold protein in assembling PAF1C46. In this study, we took yeast Paf1C as example to show that Leo1 binds to Paf1C also through Paf1 and further confirm that the Ctr9/Paf1 is the core component for Paf1C assembly and functional regulation (Figs.聽6 and 7). Although these conserved structure and function, there exist some difference between human PAF1C and yeast Paf1C. For examples, human RTF1 is a less stable subunit of PAF1C5,7,58, while the cognate yeast Rtf1 binds to Paf1C tightly (lanes 2 in Fig.聽6a, b); and the observation that the N-terminal tail of yeast Ctr9 (not human CTR9) is essential for its binding to Paf1 (Fig.聽5). Further studies are necessary to reveal the evolutional structure and function or difference of this holocomplex.TPR motifs are generally composed of 伪-helical elements and are involved in protein鈥損rotein interactions with great ligand-binding-mode diversity47. A typical example is that of an extended short-peptide binding conformation that can be observed in the Hop/Hsp90 interactions59. There, the ligand Hsp90 binds the concave side of the Hop TPR motifs. In another representative case, the ligand Caf4 forms a U-fold interacting with both the concave and convex sides of the Fis1 TPRs60. Notably, our structural analyses show that both ligands human PAF1 and yeast Paf1 formed a similar hookfold binding to both the concave channel (folded by the TPR1鈥? and TPR4鈥? subdomains of CTR9 or Ctr9) and the convex surface of TPR4鈥? (Figs.聽1, 2, 4, and Supplementary Fig.聽5g), thus defining a, to the best of our knowledge, previously uncharacterized mode of interaction among TPR motifs.Structural-based alignment indicated that the residues D79, G81, and D85 in Region I of human PAF1 are conserved from yeast to human or the residue D95 in Region II are conserved from worm to human (Fig.聽2e), indicating these residues have important roles in complexes formation. It was noted that G81 (cognate residue G63 of yeast Paf1) induces a bend that is critical for L80 and V82 (cognate residues L62 and M64 of Paf1) to fit into their respective hydrophobic pockets, respectively (Supplementary Fig.聽5a, d, h, i). Interestingly, the residues D79, D85, and D95 of PAF1 do not make contact with CTR9 (Fig.聽2b, c), which is different to the cognate residues D61 and D67 involved in binding to Ctr9 (Supplementary Fig.聽5h). However, these amino acids are very important for maintaining the local folding of PAF1. The side chain oxygen of D79 forms a hydrogen bond with the backbone nitrogen of G81; the side chain oxygen of D85 forms hydrogen bonds with the side chain hydroxyl oxygen of T91 and backbone nitrogen of I87 or N88; the side chain oxygen of D95 forms hydrogen bond with backbone nitrogen of N97 (Supplementary Fig.聽5i). So we hypothesized that the disease-associated mutation D85N or D95Y (category 2 shown in Fig.聽3a) will affect these hydrogen bonds and change the local folding of PAF1. To this end, we tested the mutants PAF1(57鈥?16)(D85N) and PAF1(57鈥?16)(D95Y) in their binding to CTR9. Size-exclusion chromatography and SDS-PAGE analysis showed the mutant PAF1(57鈥?16)(D85N) or PAF1(57鈥?16)(D95Y) can form a complex with CTR9(1鈥?49) (Supplementary Fig.聽1j and SDS-PAGEs in Supplementary Fig.聽1j1 and j2). However, a temperature-dependent denaturation assay demonstrated that both the mutant complexes CTR9(1鈥?49)/PAF1(57鈥?16)(D85N) and CTR9(1鈥?49)/PAF1(57鈥?16)(D95Y) are less stable than CTR9(1鈥?49)/PAF1(57鈥?16) WT complex (orange and magenta lines comparing to red line and inset in Supplementary Fig.聽1i).Size-exclusion chromatography and SDS-PAGE analysis showed that the purified human CTR9(1鈥?49) is aggregated in absence of PAF1(57鈥?16) binding (Supplementary Fig.聽6). These data indicate that human CTR9/PAF1 subcomplex formation is essential for the folding of CTR9 (at least for CTR9(1鈥?49)).PAF1C has been implicated in tumorigenesis3. In particular, PAF1 is involved in pancreatic and ovarian cancer32,33, and CTR9 is involved in Wilms tumor and breast cancer34,35. These findings, together with the observation that Paf1/Ctr9 heterodimer formation is essential for Paf1C assembly and function (Figs.聽6 and 7, and ref. 46), suggest that the disease-associated missense mutations (categories 1 and 2 shown in Fig.聽3a) or other mutations in CTR9 (Supplementary Table聽2) and PAF1 (Supplementary Table聽3) may affect the integrity of a functional PAF1C, which affects gene regulation (e.g., via H3H4 di- and trimethylation) in the progress of these diseases. Future studies are necessary to evaluate this hypothesis.MethodsProtein expression and purificationTo co-express a fragment of human CTR9 (residues 1鈥?84, CTR9(1鈥?84)), full-length human PAF1 (residues 1鈥?31, PAF1), and full-length human LEO1 (residues 1鈥?66, LEO1), corresponding gene fragments were amplified by polymerase chain reaction (PCR) from the cDNA library of the human kidney epithelial (HEK) 293T cell line (CBTCCAS, GNHu17). PAF1 and CTR9(1鈥?84) were cloned into two multiple cloning sites of pETDuet-1 vector (Novagen) separately and sequentially. LEO1 was cloned into an in-house modified version of the pET32a vector (Novagen). Ribosome binding sequence (RBS)-LEO1, which contains an RBS on the N-terminus of LEO1, was inserted sequentially before CTR9(1鈥?84). All of the resulting proteins contained a thioredoxin (Trx)-his6 tag on the N-terminus.The tripartite complex recombinant proteins were expressed in BL21(DE3) Codon Plus Escherichia coli at 16鈥壜癈 for 16鈥?8鈥塰. The cells were then lysed by high pressure cell cracker AH-1500 (ATS Engineering Limited). The Trx-His6-tagged protein complex was purified by Ni-NTA affinity chromatography (QIAGEN) followed by size-exclusion chromatography on a HiLoad 26/60 Superdex 200 (GE Healthcare) in 50鈥塵M Tris-HCl, pH 8.0, 200鈥塵M NaCl, 1鈥塵M EDTA, and 1鈥塵M dithiothreitol (DTT). SDS-PAGE analysis of the recombinant Trx-CTR9(1鈥?84)/Trx-PAF1/Trx-LEO1 complex is shown in Supplementary Fig.聽1a.The similar co-expression strategies were used to prepare various truncations of both the human CTR9/PAF1 and yeast Ctr9/Paf1 complexes, including CTR9(1鈥?84) and residues 57鈥?66 of PAF1 (referred to as CTR9(1鈥?84)/PAF1(57鈥?66), similar nomenclature hereafter), CTR9(1鈥?84)/PAF1(57鈥?40), CTR9(1鈥?84)/PAF1(57鈥?16), CTR9(1鈥?49)/PAF1(57鈥?16), Ctr9(1鈥?13)/Paf1(34鈥?03), and Ctr9(1鈥?70)/Paf1(34鈥?03). All of the expression and purification processes were similar to the procedures for the tripartite complex Trx-CTR9(1鈥?84)/Trx-PAF1/Trx-LEO1. The SDS-PAGE analyses of all recombinant protein complexes from human and yeast species are shown in Supplementary Fig.聽1 and Supplementary Fig.聽7, respectively.To make a single-chain fusion protein of the CTR9(1鈥?49)/PAF1(57鈥?16) and the Ctr9(1鈥?13)/Paf1(34鈥?03) (hereafter named as CTR9(1鈥?49)鈥揚AF1(57鈥?16) and Ctr9(1鈥?13)鈥揚af1(34鈥?03), respectively) complexes, DNA fragments were amplified by PCR and linked with a TEV protease-cleavable segment (Glu-Asn-Leu-Tyr-Phe-Gln-Ser). Two amino acids (Ser-Gly) were inserted both sides of the TEV protease-cleavable segment. The single open reading frame was cloned into an in-house modified version of the pET32a vector. The resulting protein contained a Trx-His6 tag on its N-terminus.The recombinant protein was expressed in BL21(DE3) Codon Plus E. coli cells at 16鈥壜癈 for 16鈥?8鈥塰. The Trx-His6-tagged protein was purified by Ni-NTA affinity chromatography, followed by size-exclusion chromatography on a HiLoad 26/60 Superdex 200 column in 50鈥塵M Tris-HCl, pH8.0, 200鈥塵M NaCl, 1鈥塵M EDTA, and 1鈥塵M DTT. After digestion with PreScission Protease to cleave the N-terminal Trx-His6 tag, the target protein was purified on a Hiprep Q FF 16/10 anion-exchange column. The final purification step was size-exclusion chromatography on a HiLoad 26/60 Superdex 200 column in 20 mM N-2-Hydroxyethylpiperazine-N0-2-ethanesulfonic acid (HEPES), pH 7.5, 150鈥塵M NaCl, 1鈥塵M EDTA, and 1鈥塵M DTT.The Se-Met-substituted protein was expressed in methionine auxotrophic E. coli B834 (DE3) cells grown in LeMaster medium. All of the recombinant proteins were purified by Ni-NTA agarose affinity chromatography followed by ion-exchange and size-exclusion chromatography.Crystallization and data collectionThe crystals of human CTR9(1鈥?49)鈥揚AF1(57鈥?16) complex were grown at 4鈥壜癈 at a protein concentration of 5鈥塵g/mL using the sitting drop vapor diffusion method. The native and the Se-Met-substituted protein were all equilibrated against a reservoir solution of 0.1鈥塎 Tris-HCl, pH 8.0, 27.5% PEG 400 for 3鈥塪. The crystals formation and growth was optimize using 27.5% 2-methyl-2,4-pentanediol as an additive. The crystals of the Se-Met-substituted yeast Ctr9(1鈥?13)鈥揚af1(34鈥?03) complex were grown at 20鈥壜癈 at a protein concentration of 30鈥塵g/mL using the same method as above. The protein was equilibrated against a reservoir solution of 0.1鈥塎 Tris-HCl, pH8.5, 0.8鈥塎 Lithium sulfate, 0.01鈥塎 Nickel chloride and for 3鈥塪. Then, we used the silver bullet additive (Hampton Research) containing 0.02鈥塎 HEPES-Na+, pH 6.8, 0.004鈥塎 CaCl2, 0.004鈥塎 MgCl2, 0.004鈥塎 MnCl2, and 0.004鈥塎 ZnCl2 for further optimization to improve crystal quality. The crystals were frozen in a cryo-protectant that consisted of the reservoir solution supplemented with 30% mixture including 50% Ethylene glycol and 25% NDSB-201. All diffraction data were collected at the Shanghai Synchrotron Radiation Facility (SSRF) on beamlines BL17U61 or BL19U using a CCD detector cooled to 100鈥塊. All datasets were processed and scaled using the HKL2000 software package62.Structure determination and refinementPhasing and initial model building of human CTR9(1鈥?49)鈥揚AF1(57鈥?16) complex crystal structure were determined by single wavelength anomalous dispersion (SAD) using PHENIX AutoSol wizard63 and AutoBuild wizard64, respectively. The initial phases and models of yeast Ctr9(1鈥?13)鈥揚af1(34鈥?03) complex were determined by SAD using the Shelx C/D/E program65. Then, the initial models were further rebuilt and adjusted manually with Coot program66 and were refined by phenix.refine program of PHENIX67. The final model was further validated using MolProbity68. The CTR9(1鈥?49)鈥揚AF1(57鈥?16) structure was refined with Ramachandran statistics of 96.1% favored, 3.9% allowed, and 0% outliers. The Ctr9(1鈥?13)鈥揚af1(34鈥?03) structure was refined with 95.2% favored, 4.8% allowed, and 0% outliers. Detailed data collection and refinement statistics are summarized in supplementary Table聽1. All structural figures were prepared using PyMOL (http://www.pymol.org/).Analytical ultracentrifugationSV experiments were performed in a Beckman Coulter XL-I analytical ultracentrifuge (Beckman Coulter) using double sector centerpieces and sapphirine windows. An additional protein purification step on a HiLoad 26/60 Superdex 200 gel filtration column in 20鈥塵M HEPES, pH 7.5, 150鈥塵M NaCl, 1鈥塵M EDTA, and 1鈥塵M DTT was performed before the experiments. SV experiments were conducted at 42,000鈥塺pm and 4鈥壜癈 using interference detection. The SV data were analyzed using the SEDFIT program69,70.Differential scanning fluorimetryDifferential scanning fluorimetry (DSF) was performed on a CFX96 real-time PCR instrument (Bio-Rad). A 20鈥壩糒 DSF solution is composed of 1鈥?鈥塵g/ml of protein and 9脳 SYPRO Orange (Invitrogen) in the assay buffer containing 50鈥塵M Tris, pH 7.5, 150鈥塵M NaCl, 1鈥塵M EDTA, and 1鈥塵M DTT. During DSF assays, all samples were heated from 4 to 65鈥壜癈 at a rate of 0.5鈥壜癈/min. Protein denaturation was monitored by the increased fluorescence signal of SYPRO Orange, which captures exposed hydrophobic residues during thermal unfolding. Fluorescence was measured using the VIC channel of the CFX96. The recorded melting curves were analyzed by the CFX Manager software (Bio-Rad). The temperature corresponding to the point of inflection was defined as the melting temperature (Tm).Cell culture and transfectionHEK293T (CBTCCAS, GNHu17) cells were cultured in DMEM (Sigma, USA) supplemented with 10% (vol/vol) FBS (Biological Industries) at 5% CO2 and 37鈥壜癈. HEK293T cells were transfected with Polyethylenimine (Polysciences, Inc.) according to the manufacturer鈥檚 protocol.Co-immunoprecipitationHEK293T cells were transfected with the indicated combinations of plasmids. After 24鈥塰 transfection, HEK293T cells were lysed using ice-cold cell lysis buffer (50鈥塵M Tris-HCl, pH 7.4, 150鈥塵M NaCl, 3% glycerol, 0.5% NP40, 0.5% Triton X-100, 1鈥塵M phenylmethylsulfonyl fluoride, and protease inhibitor cocktails) and cleared by centrifugation at 13,000鈥塺pm for 20鈥塵in at 4鈥壜癈. The supernatants were then incubated with agarose-conjugated anti-GFP antibody for 30鈥塵in at 4鈥壜癈. The agarose beads were washed three times with cell lysis buffer and eluted with SDS sample buffer. Samples were then subjected to SDS-PAGE and western blot analysis.Western blottingThe proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membrane (Millipore). The membranes were subsequently blocked with 10% nonfat milk in TBST (50鈥塵M Tris-HCl, pH 7.4, 150鈥塵M NaCl, and 0.1% Tween 20) for 1鈥塰. The PVDF membranes were immunoblotted with anti-Myc antibody (Sigma, M4439) diluted 1:5000 and anti-GFP (Sigma, G1544) antibody diluted 1:5000 at room temperature for 1鈥塰, and then probed with horseradish-peroxidase-conjugated secondary antibodies with a dilution of 1:5000 (Santa Cruz, sc-2004 and sc-2005) and developed with a chemiluminescent substrate (Millipore). Protein bands were visualized on the Tanon-5200 Chemiluminescent Imaging System (Tanon Science and Technology). All experiments were repeated at least three times.Yeast culture and growth conditionsYeast cells were grown in rich (YPD; 1% yeast extract, 2% peptone, and 2% glucose) or synthetic minimal medium (synthetic minimal media; 0.67% yeast nitrogen base, 2% glucose, and amino acids as needed) at 30 or 37鈥壜癈. In some experiments, cells were also cultured in the presence of 1鈥塎 NaCl or 100鈥塵M HU.Yeast strainsThe S. cerevisiae strains used in this study are summarized in Supplementary Table聽4. For gene disruption, the PAF1 coding regions were replaced with the S. cerevisiae URA3 using PCR primers containing 60 bases of identity to the regions flanking the open reading frame (ORF). The forward primer sequence is 5\'-GTACAATAGAACAGTGCTCATAATAGTATAAAGGGTCACAGCTTCGTACGCTGCAGGTCG-3\', and the reverse primer sequence is 5\'-CAGGTTTAAAATCAATCTCCCTTCACTTCTCAATATTCTAGCATAGGCCACTAGTGGATC-3\'.聽To construct a plasmid expressing WT PAF1 under the control of the endogenous promoter, a DNA fragment containing the PAF1 locus (including 1000 base pairs of the PAF1 5\' sequence and the PAF1 ORF) was amplified from S. cerevisiae genomic DNA. The corresponding variants Paf1(4S), Paf1(L83S), and Paf1(D95K) were constructed using site-directed mutagenesis. All of the recombinant fragments were inserted into vectors pRS405 using BamHI and XhoI sites. The resulting plasmids pP1K-Paf1(405), pP1K-Paf1(4S)(405), pP1K-Paf1(L83S)(405), and pP1K-Paf1(D95K)(405) were linearized by SnabI before transforming into paf1螖聽strain. When necessary, empty integration plasmids were transformed into appropriate yeast strains to ensure that strains being compared possessed the same auxotrophic genotype.Histone methylation in vivoYeast cells were grown at 30鈥壜癈 in YPD medium to stable log phase (OD600鈥夆増鈥?.6鈥?.0). Then, cells were harvested and washed in 1鈥塵l zymolyase buffer (50鈥塵M Tris-HCl at pH 7.5, 10鈥塵M MgCl2, and 1鈥塎 sorbitol) once. The yeast cells were resuspended with 470鈥壩糽 zymolyase buffer, 5鈥壩糽 DTT and 25鈥壩糽 zymolyase and incubated at 30鈥壜癈 for 30鈥塵in. Then, pellets were collected and resuspended with 600鈥壩糽 EBX (20鈥塵M Tris-HCl at pH 7.4, 100鈥塵M NaCl, 0.25% NP40, 0.5% Triton X-100, 50鈥塵M Na-butyrate) on ice for 15鈥塵in. The lysates were centrifuged, and the resulting pellets were resuspended with 900鈥壩糽 NIB (20鈥塵M Tris-HCl at pH 7.5, 1.2鈥塎 Sucrose, 100鈥塵M NaCl) on ice for 15鈥塵in. After discarding the supernatant, EBX containing 1% Triton X-100 was added into the pellets for another 15鈥塵in on ice. Then, the chromatin and nuclear debris were obtained. These were washed with EBX four times and added to loading buffer for SDS-PAGE, followed by immunoblotting with anti-H3 (Abcam, ab1791), H3K4me1 (Abcam, ab8895), H3K4me2 (Abcam, ab32356) and H3K4me3 (Abcam, ab8580) antibodies with dilutions of 1:2000, 1:500, 1:2000, and 1:1000, respectively. The authors declare that all data supporting the findings of this study are available within the article and its supplementary information files or form the corresponding author upon reasonable request. Coordinates of the crystal structures of human CTR9/PAF1 and yeast Ctr9/Paf1 have been deposited in the Protein Data Bank under the accession code 5ZYQ and 5ZYP, respectively. References1.Wade, P. A. et al. A novel collection of accessory factors associated with yeast RNA polymerase II. Protein Expres Purif. 8, 85鈥?0 (1996).Article聽 CAS聽Google Scholar聽 2.Jaehning, J. A. The Paf1 complex: platform or player in RNA polymerase II transcription? Biochim. Biophys. Acta 1799, 379鈥?88 (2010).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 3.Tomson, B. N. Arndt, K. M. The many roles of the conserved eukaryotic Paf1 complex in regulating transcription, histone modifications, and disease states. (BBA)鈥擥ene Regul. Mech. 1829, 116鈥?26 (2013).CAS聽Google Scholar聽 4.Krogan, N. J. et al. RNA polymerase II elongation factors of Saccharomyces cerevisiae: a targeted proteomics approach. Mol. Cell Biol. 22, 6979鈥?992 (2002).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 5.Zhu, B. et al. The human PAF complex coordinates transcription with events downstream of RNA synthesis. Genes Dev. 19, 1668鈥?673 (2005).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 6.Kim, J., Guermah, M. Roeder, R. G. The human PAF1 complex acts in chromatin transcription elongation both independently and cooperatively with SII/TFIIS. Cell 140, 491鈥?03 (2010).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 7.Rozenblatt-Rosen, O. et al. The parafibromin tumor suppressor protein is part of a human Paf1 complex. Mol. Cell Biol. 25, 612鈥?20 (2005).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 8.Squazzo, S. L. et al. The Paf1 complex physically and functionally associates with transcription elongation factors in vivo. EMBO J. 21, 1764鈥?774 (2002).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 9.Ng, H. H., Robert, F., Young, R. A. Struhl, K. Targeted recruitment of Set1 histone methylase by elongating Pol II provides a localized mark and memory of recent transcriptional activity. Mol. Cell 11, 709鈥?19 (2003).Article聽 PubMed聽 CAS聽Google Scholar聽 10.Krogan, N. J. et al. The Paf1 complex is required for histone H3 methylation by COMPASS and Dot1p: linking transcriptional elongation to histone methylation. Mol. Cell 11, 721鈥?29 (2003).Article聽 PubMed聽 CAS聽Google Scholar聽 11.Wood, A., Schneider, J., Dover, J., Johnston, M. Shilatifard, A. The Paf1 complex is essential for histone monoubiquitination by the Rad6-Bre1 complex, which signals for histone methylation by COMPASS and Dot1p. J. Biol. Chem. 278, 34739鈥?4742 (2003).Article聽 PubMed聽 CAS聽Google Scholar聽 12.Adelman, K. et al. Drosophila Paf1 modulates chromatin structure at actively transcribed genes. Mol. Cell. Biol. 26, 250鈥?60 (2006).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 13.Kim, J. et al. RAD6-mediated transcription-coupled H2B ubiquitylation directly stimulates H3K4 methylation in human cells. Cell 137, 459鈥?71 (2009).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 14.Mueller, C. L., Porter, S. E., Hoffman, M. G. Jaehning, J. A. The Paf1 complex has functions independent of actively transcribing RNA polymerase II. Mol. Cell 14, 447鈥?56 (2004).Article聽 PubMed聽 CAS聽Google Scholar聽 15.Sheldon, K. E., Mauger, D. M. Arndt, K. M. A requirement for the Saccharomyces cerevisiae Paf1 complex in snoRNA 3\' end formation. Mol. Cell 20, 225鈥?36 (2005).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 16.Penheiter, K. L., Washburn, T. M., Porter, S. E., Hoffman, M. G. Jaehning, J. A. A posttranscriptional role for the yeast Paf1-RNA polymerase II complex is revealed by identification of primary targets. Mol. Cell 20, 213鈥?23 (2005).Article聽 PubMed聽 CAS聽Google Scholar聽 17.Rozenblatt-Rosen, O. et al. The tumor suppressor Cdc73 functionally associates with CPSF and CstF 3鈥?mRNA processing factors. Proc. Natl Acad. Sci. USA 106, 755鈥?60 (2009).ADS聽 Article聽 PubMed聽 CAS聽Google Scholar聽 18.Nagaike, T. et al. Transcriptional activators enhance polyadenylation of mRNA precursors. Mol. Cell 41, 409鈥?18 (2011).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 19.Fischl, H., Howe, F. S., Furger, A. Mellor, J. Paf1 Has distinct roles in transcription elongation and differential transcript fate. Mol. Cell 65, 685鈥?98.e688 (2017).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 20.Koch, C., Wollmann, P., Dahl, M. Lottspeich, F. A role for Ctr9p and Paf1p in the regulation G1 cyclin expression in yeast. Nucleic Acids Res. 27, 2126鈥?134 (1999).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 21.Ding, L. et al. A genome-scale RNAi screen for Oct4 modulators defines a role of the Paf1 complex for embryonic stem cell identity. Cell Stem. Cell. 4, 403鈥?15 (2009).Article聽 PubMed聽 CAS聽Google Scholar聽 22.Kikuchi, I. et al. Dephosphorylated parafibromin is a transcriptional coactivator of the Wnt/Hedgehog/Notch pathways. Nat. Commun. 7, 12887 (2016).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 23.Gaillard, H. et al. Genome-wide analysis of factors affecting transcription elongation and DNA repair: a new role for PAF and Ccr4-not in transcription-coupled repair. PLoS Genet. 5, e1000364 (2009).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 24.Tatum, D., Li, W., Placer, M. Li, S. Diverse roles of RNA polymerase II-associated factor 1 complex in different subpathways of nucleotide excision repair. J. Biol. Chem. 286, 30304鈥?0313 (2011).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 25.Kowalik, K. M. et al. The Paf1 complex represses small-RNA-mediated epigenetic gene silencing. Nature 520, 248鈥?52 (2015).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 26.Portnoy, V. et al. saRNA-guided Ago2 targets the RITA complex to promoters to stimulate transcription. Cell Res. 26, 320鈥?35 (2016).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 27.Chen Fei, X. et al. PAF1, a molecular regulator of promoter-proximal pausing by RNA polymerase II. Cell 162, 1003鈥?015 (2015).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 28.Chen, F. X. et al. PAF1 regulation of promoter-proximal pause release via enhancer activation. Science 357, 1294鈥?298 (2017).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 29.Wu, L., Li, L., Zhou, B., Qin, Z. Dou, Y. H2B ubiquitylation promotes RNA Pol II processivity via PAF1 and pTEFb. Mol. Cell 54, 920鈥?31 (2014).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 30.Yu, M. et al. RNA polymerase II-associated factor 1 regulates the release and phosphorylation of paused RNA polymerase II. Science 350, 1383鈥?386 (2015).ADS聽 Article聽 PubMed聽 CAS聽Google Scholar聽 31.Lu, X. et al. Multiple P-TEFbs cooperatively regulate the release of promoter-proximally paused RNA polymerase II. Nucleic Acids Res. 44, 6853鈥?867 (2016).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 32.Moniaux, N. et al. The human homologue of the RNA polymerase II-associated factor 1 (hPaf1), localized on the 19q13 amplicon, is associated with tumorigenesis. Oncogene 25, 3247鈥?257 (2006).Article聽 PubMed聽 CAS聽Google Scholar聽 33.Vaz, A. P. et al. Novel role of pancreatic differentiation 2 in facilitating self-renewal and drug resistance of pancreatic cancer stem cells. Br. J. Cancer 111, 486鈥?96 (2014).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 34.Hanks, S. et al. Germline mutations in the PAF1 complex gene CTR9 predispose to Wilms tumour. Nat. Commun. 5, 4398 (2014).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 35.Zeng, H. Xu, W. Ctr9, a key subunit of PAFc, affects global estrogen signaling and drives ER伪-positive breast tumorigenesis. Genes Dev. 29, 2153鈥?167 (2015).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 36.Carpten, J. D. et al. HRPT2, encoding parafibromin, is mutated in hyperparathyroidism鈥搄aw tumor syndrome. Nat. Genet. 32, 676鈥?80 (2002).Article聽 PubMed聽 CAS聽Google Scholar聽 37.Chaudhary, K., Deb, S., Moniaux, N., Ponnusamy, M. P. Batra, S. K. Human RNA polymerase II-associated factor complex: dysregulation in cancer. Oncogene 26, 7499鈥?507 (2007).Article聽 PubMed聽 CAS聽Google Scholar聽 38.Amrich, C. G. et al. Cdc73 subunit of Paf1 complex contains C-terminal Ras-like domain that promotes association of Paf1 complex with chromatin. J. Biol. Chem. 287, 10863鈥?0875 (2012).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 39.Chen, H. et al. Crystallographic analysis of the conserved C-terminal domain of transcription factor Cdc73 from Saccharomyces cerevisiae reveals a GTPase-like fold. Acta Crystallogr. D 68, 953鈥?59 (2012).Article聽 PubMed聽 CAS聽Google Scholar聽 40.de Jong, R. N. et al. Structure and DNA binding of the human Rtf1 Plus3 domain. Structure 16, 149鈥?59 (2008).Article聽 PubMed聽 CAS聽Google Scholar聽 41.Wier, A. D., Mayekar, M. K., H茅roux, A., Arndt, K. M. VanDemark, A. P. Structural basis for Spt5-mediated recruitment of the Paf1 complex to chromatin. Proc. Natl Acad. Sci. USA 110, 17290鈥?7295 (2013).ADS聽 Article聽 PubMed聽Google Scholar聽 42.Mayekar, M. K., Gardner, R. G. Arndt, K. M. The recruitment of the Saccharomyces cerevisiae Paf1 complex to active genes requires a domain of Rtf1 that directly interacts with the Spt4鈥揝pt5 complex. Mol. Cell Biol. 33, 3259鈥?273 (2013).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 43.Sun, W. et al. Crystal structure of the N-terminal domain of human CDC73 and its implications for the hyperparathyroidism鈥搄aw tumor (HPT鈥揓T) syndrome. Sci. Rep. 7, 15638 (2017).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 44.Van Oss, S. B. et al. The histone modification domain of Paf1 complex subunit Rtf1 directly stimulates H2B ubiquitylation through an interaction with Rad6. Mol. Cell 64, 815鈥?25 (2016).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 45.Xu, Y. et al. Architecture of the RNA polymerase II-Paf1C-TFIIS transcription elongation complex. Nat. Commun. 8, 15741 (2017).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 46.Chu, X. et al. Structural insights into Paf1 complex assembly and histone binding. Nucleic Acids Res. 41, 10619鈥?0629 (2013).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 47.Zeytuni, N. Zarivach, R. Structural and functional discussion of the tetra-trico-peptide repeat, a protein interaction module. Structure 20, 397鈥?05 (2012).Article聽 PubMed聽 CAS聽Google Scholar聽 48.Zhang, M. et al. The noncanonical tetratricopeptide repeat (TPR) domain of fluorescent (FLU) mediates complex formation with glutamyl-tRNA reductase. J. Biol. Chem. 290, 17559鈥?7565 (2015).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 49.Jinek, M. et al. The superhelical TPR-repeat domain of O-linked GlcNAc transferase exhibits structural similarities to importin alpha. Nat. Struct. Mol. Biol. 11, 1001鈥?007 (2004).Article聽 PubMed聽 CAS聽Google Scholar聽 50.Wang, J., Dye, B. T., Rajashankar, K. R., Kurinov, I. Schulman, B. A. Insights into anaphase promoting complex TPR subdomain assembly from a CDC26鈥揂PC6 structure. Nat. Struct. Mol. Biol. 16, 987鈥?89 (2009).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 51.Zhang, Z., Kulkarni, K., Hanrahan, S. J., Thompson, A. J. Barford, D. The APC/C subunit Cdc16/Cut9 is a contiguous tetratricopeptide repeat superhelix with a homo-dimer interface similar to Cdc27. EMBO J. 29, 3733鈥?744 (2010).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 52.Zhu, J. et al. LGN/mInsc and LGN/NuMA complex structures suggest distinct functions in asymmetric cell division for the Par3/mInsc/LGN and Galphai/LGN/NuMA pathways. Mol. Cell 43, 418鈥?31 (2011).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 53.Hinshaw, S. M., Makrantoni, V., Harrison, S. C. Marston, A. L. The kinetochore receptor for the cohesin loading complex. Cell 171, 72鈥?4.e13 (2017).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 54.Mueller, C. L. Jaehning, J. A. Ctr9, Rtf1, and Leo1 are components of the Paf1/RNA polymerase II complex. Mol. Cell Biol. 22, 1971鈥?980 (2002).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 55.Betz, J. L. et al. Phenotypic analysis of Paf1/RNA polymerase II complex mutations reveals connections to cell cycle regulation, protein synthesis, and lipid and nucleic acid metabolism. Mol. Genet. Genom. 268, 272鈥?85 (2002).Article聽 CAS聽Google Scholar聽 56.Shi, X. et al. Paf1p, an RNA polymerase II-associated factor in Saccharomyces cerevisiae, may have both positive and negative roles in transcription. Mol. Cell Biol. 16, 669鈥?76 (1996).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 57.Schneider, J. et al. Molecular regulation of histone H3 trimethylation by COMPASS and the regulation of gene expression. Mol. Cell 19, 849鈥?56 (2005).Article聽 PubMed聽 CAS聽Google Scholar聽 58.Yart, A. et al. The HRPT2 tumor suppressor gene product parafibromin associates with human PAF1 and RNA polymerase II. Mol. Cell Biol. 25, 5052鈥?060 (2005).Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 59.Scheufler, C. et al. Structure of TPR domain-peptide complexes: critical elements in the assembly of the Hsp70鈥揌sp90 multichaperone machine. Cell 101, 199鈥?10 (2000).Article聽 PubMed聽 CAS聽Google Scholar聽 60.Zhang, Y. Chan, D. C. Structural basis for recruitment of mitochondrial fission complexes by Fis1. Proc. Natl Acad. Sci. USA 104, 18526鈥?8530 (2007).ADS聽 Article聽 PubMed聽Google Scholar聽 61.Wang, Q.-S. et al. Upgrade of macromolecular crystallography beamline BL17U1 at SSRF. Nucl. Sci. Tech. 29, 68 (2018).Article聽Google Scholar聽 62.Otwinowski, Z. Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Method. Enzymol. 276, 307鈥?26 (1997).Article聽 CAS聽Google Scholar聽 63.Terwilliger, T. C. et al. Decision-making in structure solution using Bayesian estimates of map quality: the PHENIX AutoSol wizard. Acta Crystallogr. D 65, 582鈥?01 (2009).Article聽 PubMed聽 CAS聽Google Scholar聽 64.Terwilliger, T. C. et al. Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard. Acta Crystallogr. D 64, 61鈥?9 (2008).Article聽 PubMed聽 CAS聽Google Scholar聽 65.Schneider, T. R. Sheldrick, G. M. Substructure solution with SHELXD. Acta Crystallogr. D 58, 1772鈥?779 (2002).Article聽 PubMed聽 CAS聽Google Scholar聽 66.Emsley, P., Lohkamp, B., Scott, W. G. Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486鈥?01 (2010).Article聽 PubMed聽 CAS聽 PubMed Central聽Google Scholar聽 67.Adams, P. D. et al. PHENIX: a comprehensive python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213鈥?21 (2010).Article聽 PubMed聽 CAS聽 PubMed Central聽Google Scholar聽 68.Davis, I. W. et al. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 35, W375鈥揥383 (2007).ADS聽 Article聽 PubMed聽 PubMed Central聽Google Scholar聽 69.Schuck, P. Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys. J. 78, 1606鈥?619 (2000).ADS聽 Article聽 PubMed聽 PubMed Central聽 CAS聽Google Scholar聽 70.Schuck, P. On the analysis of protein self-association by sedimentation velocity analytical ultracentrifugation. Anal. Biochem. 320, 104鈥?24 (2003).Article聽 PubMed聽 CAS聽Google Scholar聽 Download referencesAcknowledgmentsWe are grateful to the staff at the beamlines BL17U1 and BL19U1 of the Shanghai Synchrotron Radiation Facility (SSRF) for excellent technical assistance during data collection. We thank Dr. Jianchao Li (HKUST, Hong Kong) for his help in structure determination. This work was supported by the 973 Program 2014CB910201 (to J.L.), National Natural Science Foundation of China [31470755 and 31670758 (to J.L.); 31870750 to (to H. Z.)].Author informationAuthor notesThese authors contributed equally: Ying Xie, Minying Zheng.AffiliationsState Key Laboratory of Medicinal Chemical Biology, Tianjin Key Laboratory of Protein Science, and College of Life Sciences, Nankai University, 94 Weijin Road, Tianjin, 300071, ChinaYing Xie,聽Minying Zheng,聽Yue Chen,聽Huisha Xu,聽Hao Zhou聽 聽Jiafu LongDepartment of Epidemiology and Biostatistics, Tianjin Medical University Cancer Institute and Hospital, Tianjin, 300060, ChinaXinlei ChuState Key Laboratory of Membrane Biology, School of Life Sciences, Tsinghua University, Beijing, 100084, ChinaJiawei WangAuthorsYing XieView author publicationsYou can also search for this author in PubMed聽Google ScholarMinying ZhengView author publicationsYou can also search for this author in PubMed聽Google ScholarXinlei ChuView author publicationsYou can also search for this author in PubMed聽Google ScholarYue ChenView author publicationsYou can also search for this author in PubMed聽Google ScholarHuisha XuView author publicationsYou can also search for this author in PubMed聽Google ScholarJiawei WangView author publicationsYou can also search for this author in PubMed聽Google ScholarHao ZhouView author publicationsYou can also search for this author in PubMed聽Google ScholarJiafu LongView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsY.X., M.Z., H.Z, and J.L. designed research. Y.X., M.Z., X.C., Y.C., H.X., J.W., and H.Z. performed research. Y.X., M.Z., H.Z., and J.L. analyzed the data and prepared the figures. Y.X., M.Z., H.Z., and J.L. wrote the paper. All authors reviewed the manuscript. J.L. coordinate the research.Corresponding authorsCorrespondence to Hao Zhou or Jiafu Long.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPublisher\'s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Electronic supplementary material Seychelle M. Vos, Lucas Farnung, Andreas Linden, Henning Urlaub Patrick Cramer Nature Structural Molecular Biology (2020) Lindsey D. Goodman, Mercedes Prudencio, Nicholas J. Kramer, Luis F. Martinez-Ramirez, Ananth R. Srinivasan, Matthews Lan, Michael J. Parisi, Yongqing Zhu, Jeannie Chew, Casey N. Cook, Amit Berson, Aaron D. Gitler, Leonard Petrucelli Nancy M. Bonini Nature Neuroscience (2019) CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.