Paf1 and Ctr9 subcomplex formation is essential for Paf1 complex assembly and functional regulation

The 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.

-It is unclear how the authors built their structures. Lines 471-472: do the authors use the models generated by AutoSol/Autobuild as initial models in Coot or do they build the structure de novo into the electron density? This should be clearly described.
-It is not clear whether CTR9 1-249 is folded/well behaved on its own. If not, this would explain the size exclusion results presented in Fig S6. Lack of interaction with PAF1 results in poor solubility/behaviour of CTR9. It appears that most PAF1 mutants made with CTR9 elute in the void volume of the column. If this is true, the authors should state this in the text. Labels should not lie above the gel images.
-Human protein names should be written in all caps ie PAF1, LEO1, etc. Yeast names are written as Leo1, Paf1, etc. This will simplify the text and follows established naming conventions.
-Authors used MS and limited proteolysis to identify interacting regions of PAF1 and CTR9 (lines 97-98). Both procedures should be described in the methods section.
- Figure 4A: Leo1 appears to be weakly associated with all complexes the authors immunoprecipitate. The Leo1 levels in the total lysate appear normal. Is there an explanation for this? The figure legend for the bottom panel should state that the blot is representative of 3% of the input used for the IP, or something to this effect. Please add a short statement on your interpretion of this.
- Figure 4B: There is an additional band in the Myc blot, under LEO1. Do the authors know the identity of the band? This is not essential but maybe interesting.
- Figure S1: The nomenclature for heterodimers and fusion proteins is confusing.
-Line 96: change phrase to "eluted from size exclusion as a tripartite complex" or something similar -line 141 "TRP" should be "TPR" -line 238 "Se-substituted" should be defined. It appears that the authors used Se-Met for phase determination, and this language should also be used in the text.
-line 411: the authors should state what the AH-1500 does. Is it a sonicator, French press, etc.
-Line 430: It is unclear if the Ser-Gly additions are found at the N-or C-terminal side of the TEV site -Line 457: state as "the silver bullet additive" instead of "the silver bullet" and state manufacturer -Line 494: ºC appears to have been replaced by a box in a number of positions.
Reviewer #2 (Remarks to the Author): This manuscript describes a structure-function analysis of the association between the Paf1 and Ctr9 subunits of the Paf1 complex (Paf1C). Paf1C plays an important, albeit not fully understood, role in transcription elongation and connecting elongation to chromatin modifications, particularly activating marks on histone tails. Paf1C is critical for viability in yeast and has important connections with human disease. For example, some of the subunits and their interactions are thought to have tumor suppressor activity. There are few structural details about how the complex assembles and how the complex interacts with transcriptional machinery.
The authors have solved novel high-resolution crystal structures of two similar complexes, from human and yeast, containing the primary interacting regions of Paf1 and Ctr9. They compliment the structure with biochemical interaction studies in human cells and functional studies in yeast that exploit structure-based mutations. The main significant findings are the structural details of the Paf1-Ctr9 association, the mapping of several cancer-derived mutations to the interface, and the functional studies that demonstrate the Paf1-Ctr9 interface is critical for Paf1C assembly and function. The technical quality of the crystal structures is high and conclusions are properly supported by the complimentary biochemical data. The study is for the most part rigorously done and well explained. Considering the importance of the complex and lack of structural details currently available, the work will have impact. It should be published after the authors are given the opportunity to consider the following minor concerns and suggestions: 1) It would be useful if the authors could explain more how they envision this subcomplex fitting into the full Paf1C complex. Although the previously published low resolution EM map may not allow for unambiguous docking of the high resolution structure, the authors could at least comment on different relationships between the subunits within the complex and perhaps show a schematic model. For example, the Ctr9 binding site in Paf1 is near the Leo1 binding site. Does Leo1 also contact Ctr9? More broadly, how is the observation that this local interaction is required for overall complex assembly consistent with a model for architecture?
2) Line 84: The Ctr9-Caf1 heterodimer is called an "obligate" heterodimer, but it is not made clear what evidence supports this description. Was this published elsewhere? Fig S6 suggests that Ctr9 is aggregated in the absence of Paf1 association, but this part of the observation is not given any significant attention.
3) Lines 105-106: 1:1 stoichiometry cannot be rigorously determined from size-exclusion column (SE), particularly without standards. However, the 1:1 stoichiometry is convincingly shown by the molecular weight measurement using analytical ultracentrifugation data (and crystal structure). It would be more justified to change the conclusion from the SEC data to focus on the co-elution and the fact that the heterodimer and fusion elute in the same place, which supports a similar folded structure. 4) Line 141: Change "five TRP" to "five TPR".
5) It is curious that the A helix in Paf1 does not make contact with Ctr9 or anything that is clear in Fig. 2, yet it is ordered and somehow stabilized as a helix. Furthermore, there are some highly conserved residues and a disease related mutation in this helix. Does this helix contact a different Ctr9 in the crystal lattice that is related by symmetry? If so, this begs the questions as to whether it could also make the equivalent interaction in solution. Does deletion of the A helix not change heterodimer stability? Do the authors have other thoughts on the role of this helix?
6) The authors could comment on the role of some of the most conserved residues in Paf1 Region I that do not appear to make contact with Ctr9 (D79, G81, and D85). For example, it appears in the view in Fig. S5A that G81 induces a bend that is critical for both L80 and V82 to fit into their respective hydrophobic pockets. 7) If possible, it would be a more impactful experiment if the co-IP in HEK cells were performed with full-length proteins. It is not surprising that the structure-derived mutations show the predicted effects when using identical constructs to those in the crystals. The more important question is whether the structural results are applicable in the context of the full-length proteins.
Reviewer #3 (Remarks to the Author): In the manuscript entitled "Paf1/Ctr9 subcomplex formation is essential for Paf1 complex assembly and functional regulation" Xie et al. present the structure of a subcomplex of the human and yeast Ctr9/Paf1. This subcomplex is part of the polymerase-associated factor 1 complex (PAF1C), which is involved in transcriptional elongation and chromatin regulation. There is limited structural information about PAF1C. This work builds up on previous work (Chu et al., 2013;Xu et al., 2017) and constitutes a step forward in our understanding of the complex formation at the molecular level. Furthermore, by using a combination of biophysical and biochemistry methods, the authors demonstrate that Ctr9/Paf1 is required for efficient assembly of PAF1C. In addition to the new structural data, the authors perform key mutations on the interface of Ctr9/Paf1 and monitor complex assembly and functionality in vivo assays. These point mutations confirm the impact that the correct assembly of PAF1C might have on cancer progression, since some of the previously identified mutations in cancer disease play also a key role in complex formation.
Overall, the science, methods and statistics presented in this work appear sound. There are no major issues, however some further clarification is needed on the points given below. Additionally, the paper would be easier to understand if the presentation was improved, e.g. simplification of figures.
1. Quality of Figures 4A, 4B and 5E is poor. Replacement with higher resolution images is a necessity. 2. The authors perform co-immunoprecipitations (co-IP) experiments to test the interaction between hCtr9 and hPaf157-116. In Figure 2F, these assays are shown. In light of the crystal structure (hCtr91-249 -hPaf157-116), an extra co-IP assay showing the result of the interaction between hCtr91-249 and hPaf157-116 would complete Figure 2F (similar to lane 2 in Figure 2H). Alternatively, the authors should mention Figures 2F and 2H together whenever they refer to this experiment. 3. The authors attempt to define a new mode of interaction among TPR motifs. TPR motifs are widespread in nature and present a common fold. Further evidence would be needed to conclude that a novel TPR-SLIM (small linear motif) interaction has been found. This point seems overinterpreted. Comprehensive bioinformatics analysis and a figure showing a comparison of structural results (and different type of folds) would be necessary to raise this conclusion. Otherwise, the authors are encouraged to moderate (or even omit) this statement in the text. 4. To validate the interactions observed in the structure of hCtr91-249/hPaf157-116 a number of mutagenesis studies are presented. The overall conclusion of this study seems precise. However, Figure S6 and related comments need to be improved for clarity. For example, all mutants (except hCtr91-249 (R98W)/hPaf157-116, Figure S6C, blue lane) seem to aggregate. This is a typical behaviour of disrupted complexes that occurs when one or more of the interactor partners are missing. In Figures S6A and S6A2, Ctr9 seems to appear in the void volume (first peak of SEC, S6A) and Paf1 shows up later (second peak of SEC, S6A). In the same figure, it also seems like Ctr9 is also present in this second peak ( Figure S6A2, lane 35) and shows the same intensity as Paf1. This does not correlate with the absorbance of the chromatogram peaks. In addition, in S6A2 there are 2 bands of similar size (14.4 kDa) and it is unclear which one corresponds to hPaf157-116(5M). Could the authors elaborate further on the interpretation of the chromatograms and corresponding gels, and specifically on Figures S6A and S6A2? Could it be that mutated complexes are still observed in the second peak of the chromatograms (~fraction 35) but with disrupted stoichiometry? See comments on how to improve Figure S6A below. 5. Lines 85-86 "elucidated the mechanism responsible for human diseases caused by mutations of Paf1 and Ctr9" are an overstatement of the conclusions. The authors have demonstrated that the assembly of PAF1C might have an impact on specific cancer diseases, but further investigation is needed to elucidate the mechanism responsible for the human disease.
Suggestions to improve figures.   Figure 2E. The name of the proteins used to depict secondary structures on top (hPaf1) and bottom (yPaf1) of alignment should be indicated for clarity. Have these secondary structures been predicted with any specific program? The same applies to Figure S2 (hCtr9 and yCtr9). • Figures 2F and 2H. Labels of specific band proteins are not shown in Figures 2F and 2H. Also, is hPaf157-116 tagged with GFP in N or C terminus? (N-terminal residue in Figure 2F and C-terminal one in Figure 2H). • Figure 2G. To make it easier understand what type or residues are included in categories 1, 2, 3 (both mentioned in the text and in this figure), these could be renamed as 'interface', 'folding' and 'others', respectively. This figure is difficult to interpret and it is not very informative (again a comprehensive table with mutations would help if added). To make this figure more useful, the authors could focus on showing just the mutations used in this work (categories 1 and 3). Figure 5E. Label of IB is missing. Figure S1. • Contrast of S1B needs to be improved. • Color code for chromatograms is not shown in S1E. Figure S3. This figure can be used to highlight the differences in the N-terminal domain of the human and yeast proteins, which will complement Figure 3H. This represents an important finding in this paper, i.e. this N-terminal region is essential to maintain the interaction between yCtr9 and yPaf1. Figure S4C. To appreciate the differences between human and yeast Ctr9/Paf1, the use of only two colours would simplify the representation. Figure S5. • S5A and S5B. It is not possible to appreciate the stick representation of residues very well. A zoom highlighting these regions would help the reader. Labels in green and blue are not easy to read. A possibility could be to use lighter colours. • Figure S5E. Not mention in the text. Minor points.
-Several parts of the manuscript need to be clarified so it is more suitable for a general audience. This will aid the reader understand better the text. For example, a general overview of what it is known about Paf1, Ctr9 and PAF1C from the structural point of view is missing.
-'Crystal structure of the hCtr9/hPaf1 heterodimer' section. It is unclear how the TEV constructs have been made. Is the TEV site cleaved at all? Specify in line 117 where the TEV-cleavable segment has been included. Explain how the artificial linker is helping crystallization. -Lane 267. Elaborate further on the interaction between the N-terminal tail of yCtr9 and yPaf1 and how this is different to the human system. -Lane 268. Specify that yL83 is tested because it is only present in yeast.
-Different terms are used for co-immunoprecipitation (co-IP). For example in lines 290 and 294 it is named as coprecipitation. Check consistency for this term.
-A cartoon showing the updated proposed assembly of PAF1C is recommended. This would enhance the result and discussion. -D95K mutation in yPaf1 abolished the interaction between yCtr9 and yPaf1 completely. It would be interesting to test if D104K in hPaf1 had the same effect.

(Our responses and all changes in the revised manuscript are in blue)
Reviewer #1 (Remarks to the Author): The authors report high-quality crystallographic structures of the N-terminal region of Ctr9 bound to a N-terminal tether in the Paf1 subunit. Our structural understanding of the Paf1 complex is still very fragmented and therefore these structures, for the yeast and human complexes, are important. The authors add some functional data to demonstrate the relevance of the observed interaction. This reviewer supports rapid publication of these competitive structures in Nature Communications after the following minor concerns have been addressed.
We thank the reviewer for supporting to publish our work in Nature Communications, and for the constructive critiques and suggestions below.
-It is unclear how the authors built their structures. Lines 471-472: do the authors use the models generated by AutoSol/Autobuild as initial models in Coot or do they build the structure de novo into the electron density? This should be clearly described.
Thanks for pointing out this unclear description. Phasing and initial model building of human CTR9 (1-249) -PAF1 (57-116) complex crystal structure were determined by single wavelength anomalous dispersion (SAD) using PHENIX AutoSol wizard and AutoBuild wizard, respectively. The initial phases and models of yeast Ctr9 (1-313) -Paf1 (34-103) complex were determined by SAD using the Shelx C/D/E program. Then, the initial models were further rebuilt and adjusted manually with Coot program and were refined by phenix.refine program of PHENIX. The final model was further validated using MolProbity. We have modified the method accordingly.
-It is not clear whether CTR9 1-249 is folded/well behaved on its own. If not, this would explain the size exclusion results presented in Fig S6. Lack of interaction with PAF1 results in poor solubility/behaviour of CTR9. It appears that most PAF1 mutants made with CTR9 elute in the void volume of the column. If this is true, the authors should state this in the text. Labels should not lie above the gel images.
Agree with the reviewer's comments. The purified human CTR9 (1-249) recombinant protein was aggregated in size-exclusion column (blue line in Fig. I, below for convenience), comparing to CTR9 (1-249) /PAF1 (57-116) complex protein (black line in Fig. I). Additionally, we found that yeast Ctr9 (1-313) forms inclusion bodies when expressing in E. coli (data now shown). These data indicate that both human CTR9/PAF1 and yeast Ctr9/Paf1 subcomplexes formation are essential for the folding of CTR9 and Ctr9 (at least for CTR9  and Ctr9  ), respectively. We have included  -Human protein names should be written in all caps ie PAF1, LEO1, etc. Yeast names are written as Leo1, Paf1, etc. This will simplify the text and follows established naming conventions.
Thanks for suggestion. Following the reviewer's suggestion, we have modified the manuscript accordingly. Additionally, complex names of human and yeast species are named as PAF1C and Paf1C, respectively.
-Authors used MS and limited proteolysis to identify interacting regions of PAF1 and CTR9 (lines 97-98). Both procedures should be described in the methods section.
Thanks for suggestion. Actually, after finishing limited proteolysis using trypsin, we found that the Mass spectrometer was off service. At that time, according to the results of limited proteolysis, together with our previous results shown in Ref. #46 in this manuscript and the sequence alignment of PAF1 from various species, we tested the region in PAF1 (aa 57-266) whether can bind to CTR9  or not independent of the LEO1 subunit. Fortunately, we got the positive result. We have modified the manuscript accordingly.
- Figure 4A: Leo1 appears to be weakly associated with all complexes the authors immunoprecipitate. The Leo1 levels in the total lysate appear normal. Is there an explanation for this? The figure legend for the bottom panel should state that the blot is representative of 3% of the input used for the IP, or something to this effect. Please add a short statement on your interpretion of this.
We agree with the reviewer on this comment and thank the reviewer for pointing this issue out. To figure this query out ,we have optimized the co-IP experiments and repeated at least three times (representative data in Fig. II  - Figure 4B: There is an additional band in the Myc blot, under LEO1. Do the authors know the identity of the band? This is not essential but maybe interesting.
Thanks for pointing this issue out. To answer this question, we have examined the identity of the additional band via a co-IP strategy and confirm that this band is the degradation product of Ctr9, since this band will appear in the context of co-IP assays adding Ctr9 (  - Figure S1: The nomenclature for heterodimers and fusion proteins is confusing. "/" denotes protein complexes with separate chains (e.g., CTR9 (1-249) /PAF1 (57-116) ), while "-" denotes heterodimers in a single-chain fusion (e.g., CTR9 (1-249) -PAF1 (57-116) ). We have modified manuscript and legend accordingly.
-Line 96: change phrase to "eluted from size exclusion as a tripartite complex" or something similar According to the reviewer's suggestion, we have revised the phrase to "eluted from the size-exclusion column as a tripartite complex".
-line 141 "TRP" should be "TPR" Thank you for noticing this mistake. We have corrected this error in the revised manuscript.
-line 238 "Se-substituted" should be defined. It appears that the authors used Se-Met for phase determination, and this language should also be used in the text.
We have corrected this inconsistent description in the revised manuscript.
-line 411: the authors should state what the AH-1500 does. Is it a sonicator, French press, etc.
The AH-1500 is a type of high pressure cell cracker. We have included this information in the methods section.
-Line 430: It is unclear if the Ser-Gly additions are found at the N-or C-terminal side of the TEV site DNA fragments were amplified by PCR and linked with a tobacco etch virus (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. We have corrected this unclear description in the methods section.
-Line 457: state as "the silver bullet additive" instead of "the silver bullet" and state manufacturer Following the reviewer's suggestion, we have modified the manuscript accordingly.
-Line 494: ºC appears to have been replaced by a box in a number of positions.
We have corrected the error accordingly in the revised manuscript.
Reviewer #2 (Remarks to the Author): This manuscript describes a structure-function analysis of the association between the Paf1 and Ctr9 subunits of the Paf1 complex (Paf1C). Paf1C plays an important, albeit not fully understood, role in transcription elongation and connecting elongation to chromatin modifications, particularly activating marks on histone tails. Paf1C is critical for viability in yeast and has important connections with human disease. For example, some of the subunits and their interactions are thought to have tumor suppressor activity. There are few structural details about how the complex assembles and how the complex interacts with transcriptional machinery.
The authors have solved novel high-resolution crystal structures of two similar complexes, from human and yeast, containing the primary interacting regions of Paf1 and Ctr9. They compliment the structure with biochemical interaction studies in human cells and functional studies in yeast that exploit structure-based mutations. The main significant findings are the structural details of the Paf1-Ctr9 association, the mapping of several cancer-derived mutations to the interface, and the functional studies that demonstrate the Paf1-Ctr9 interface is critical for Paf1C assembly and function. The technical quality of the crystal structures is high and conclusions are properly supported by the complimentary biochemical data. The study is for the most part rigorously done and well explained. Considering the importance of the complex and lack of structural details currently available, the work will have impact. It should be published after the authors are given the opportunity to consider the following minor concerns and suggestions: We thank the reviewer for nicely summarizing the key contributions of our work, and for the constructive critiques and suggestions below.
1) It would be useful if the authors could explain more how they envision this subcomplex fitting into the full Paf1C complex. Although the previously published low resolution EM map may not allow for unambiguous docking of the high resolution structure, the authors could at least comment on different relationships between the subunits within the complex and perhaps show a schematic model. For example, the Ctr9 binding site in Paf1 is near the Leo1 binding site. Does Leo1 also contact Ctr9? More broadly, how is the observation that this local interaction is required for overall complex assembly consistent with a model for architecture?
We appreciate the reviewer pointing out some very important concern out. During revision , we have repeated co-IP experiments about yeast Paf1C assembly (revised Fig. 5a and 5b, and  2) Line 84: The Ctr9-Caf1 heterodimer is called an "obligate" heterodimer, but it is not made clear what evidence supports this description. Was this published elsewhere? Fig S6 suggests that Ctr9 is aggregated in the absence of Paf1 association, but this part of the observation is not given any significant attention.
We have deleted the word "obligate" in the revised manuscript. Agree the reviewer's comments, the purified human CTR9 (1-249) recombinant protein was aggregated in size-exclusion column (blue line in Fig. V, below), comparing to CTR9 (1-249) /PAF1  complex protein (black line in Fig. V). Additionally, we found that yeast Ctr9 (1-313) forms inclusion bodies when expressing in E. coli (data now shown). These data indicate that both human CTR9/PAF1 and yeast Ctr9/Paf1 subcomplexes formation are essential for the folding of CTR9 and Ctr9 (at least for CTR9  and Ctr9 (1-313) ), respectively. We have included Fig.  V as the revised Supplementary Fig. 6a and a1-a3 and modified the sections of results and discussion accordingly. 3) Lines 105-106: 1:1 stoichiometry cannot be rigorously determined from size-exclusion column (SE), particularly without standards. However, the 1:1 stoichiometry is convincingly shown by the molecular weight measurement using analytical ultracentrifugation data (and crystal structure). It would be more justified to change the conclusion from the SEC data to focus on the co-elution and the fact that the heterodimer and fusion elute in the same place, which supports a similar folded structure.
We fully agree with the reviewer on these comments. We have modified the manuscript accordingly. 4) Line 141: Change "five TRP" to "five TPR".
Thank you for noticing this mistake. We have corrected this error in the revised manuscript.
5) It is curious that the A helix in Paf1 does not make contact with Ctr9 or anything that is clear in Fig. 2, yet it is ordered and somehow stabilized as a helix. Furthermore, there are some highly conserved residues and a disease related mutation in this helix. Does this helix contact a different Ctr9 in the crystal lattice that is related by symmetry? If so, this begs the questions as to whether it could also make the equivalent interaction in solution. Does deletion of the A helix not change heterodimer stability? Do the authors have other thoughts on the role of this helix?
We thank the reviewer for pointing this very important concern out. We have analyzed two structures and found that both A of PAF1  and A' of Paf1 (34-103) do not contact with a different CTR9 (1-249) and Ctr9  , respectively, in the crystal lattice. To further evaluate the role of this N-terminal helix, a truncated fragment of human PAF1 (aa 75-116, PAF1 (75-116) ) or a truncated fragment of yeast Paf1 (aa 59-103, Paf1 (59-103) ) was designed to delete the Nterminal helix. As expected, size-exclusion chromatography and SDS-PAGE analysis showed the truncated PAF1 (75-116) still can form a complex with CTR9 (1-249) (Fig. VIa, below, insert), which is consistent with the observation that A of PAF1 (57-116) is not important for complex formation of CTR9 (1-249) -PAF1 (57-116) (the revised Figs. 1 and 2). However, a temperaturedependent denaturation assay demonstrated that the truncated CTR9 (1-249) /PAF1  complex is less stable than CTR9 (1-249) /PAF1 (57-116) complex (Fig. VIb, below), indicating that the A of PAF1 (57-116) may contribute to the stability of the CTR9 (1-249) /PAF1 (57-116) complex. We have gotten similar conclusion about the role of A' of Paf1  in yeast Ctr9 (1-313) /Paf1 (34-103) complex formation, according the results shown in the revised Supplementary Fig. 7b, b2 and 7d. We have included Fig. VI as the revised Supplementary Fig. 1h and 1i and modified the manuscript accordingly. 6) The authors could comment on the role of some of the most conserved residues in Paf1 Region I that do not appear to make contact with Ctr9 (D79, G81, and D85). For example, it appears in the view in Fig. S5A that G81 induces a bend that is critical for both L80 and V82 to fit into their respective hydrophobic pockets.
Thanks for the reviewer's suggestion. 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. 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 (revised Supplementary Fig. 5a, d, h, and i). Interestingly, the residues D79, D85, and D95 of PAF1 do not make contact with CTR9 (revised Fig. 2b and 2c), which is different to the cognate residues D61 and D67 involved in binding to Ctr9 (Fig. VIIa). 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 (Fig. VIIb, below). So we hypothesized that the disease-associated mutation D85N or D95Y will affect these hydrogen bonds and change the local folding of PAF1 (revised

7)
If possible, it would be a more impactful experiment if the co-IP in HEK cells were performed with full-length proteins. It is not surprising that the structure-derived mutations show the predicted effects when using identical constructs to those in the crystals. The more important question is whether the structural results are applicable in the context of the fulllength proteins.
We agree the reviewer on this important point. During mapping the minimal binding region of interaction between human CTR9 and PAF1, we have designed a mutant lacking the N-terminal amino acids 1-116 (referred to as PAF1(N116) and showed that the mutant GFP-[PAF1(N116)] did not co-immunoprecipitate with Myc-CTR9 thoroughly (lane 3 in Fig. VIII, below), indicating that the N-terminal fragment including amino acids 57-116 (aa 57-116 used in this study for structure determination) is required for the interaction between full-length CTR9 and PAF1. However, we found that CTR9 lacking amino acids 1-249 (aa 1-249 used in this study for structure determination) was not expressed in HEK293T cells.
Our previous study indicated that LEO1 binds to human PAF1C through PAF1 and that the CTR9 subunit is the key scaffold protein in assembling PAF1C (Ref. #46 in this manuscript). 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. 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 PAF1C (Refs. #5, #7, and #58 in this manuscript), while the cognate yeast Rtf1 binds to Paf1C tightly (lanes 2 in Fig. 5a and 5b); and the observation that the Nterminal tail of yeast Ctr9 (not human CTR9) is essential for its binding to Paf1 (revised Fig.  4). Further studies are necessary to reveal the evolutional structure and function or difference of this holo-complex. Over the past several years, we have tried very hard to obtain the holocomplex protein suitable for structural characterization and made some progress. Nevertheless, the in vitro and in vivo data presented in this manuscript clearly indicated that Ctr9/Paf1 subcomplex is the core component and is essential for holo-complex assembly. We have included Fig. VIII as the revised Supplementary Fig. 1g and modified the manuscript accordingly.

Fig. VIII. The N-terminal fragment amino acids 1-116 of PAF1 is required for the interaction between full-length CTR9 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.
Reviewer #3 (Remarks to the Author): In the manuscript entitled "Paf1/Ctr9 subcomplex formation is essential for Paf1 complex assembly and functional regulation" Xie et al. present the structure of a subcomplex of the human and yeast Ctr9/Paf1. This subcomplex is part of the polymerase-associated factor 1 complex (PAF1C), which is involved in transcriptional elongation and chromatin regulation. There is limited structural information about PAF1C. This work builds up on previous work (Chu et al., 2013;Xu et al., 2017) and constitutes a step forward in our understanding of the complex formation at the molecular level. Furthermore, by using a combination of biophysical and biochemistry methods, the authors demonstrate that Ctr9/Paf1 is required for efficient assembly of PAF1C. In addition to the new structural data, the authors perform key mutations on the interface of Ctr9/Paf1 and monitor complex assembly and functionality in vivo assays. These point mutations confirm the impact that the correct assembly of PAF1C might have on cancer progression, since some of the previously identified mutations in cancer disease play also a key role in complex formation.
Overall, the science, methods and statistics presented in this work appear sound. There are no major issues, however some further clarification is needed on the points given below. Additionally, the paper would be easier to understand if the presentation was improved, e.g. simplification of figures.
We thank the reviewer for nicely summarizing the key findings of our work, and for the constructive critiques and suggestions below, especially in improving figure presentation. Figures 4A, 4B and 5E is poor. Replacement with higher resolution images is a necessity.

Quality of
We have used higher resolution images in the revised Figs. 4a, 4b, and 6e. 2. The authors perform co-immunoprecipitations (co-IP) experiments to test the interaction between hCtr9 and hPaf157-116. In Figure 2F, these assays are shown. In light of the crystal structure (hCtr91-249 -hPaf157-116), an extra co-IP assay showing the result of the interaction between hCtr91-249 and hPaf157-116 would complete Figure 2F (similar to lane 2 in Figure 2H). Alternatively, the authors should mention Figures 2F and 2H together whenever they refer to this experiment.
Following the reviewer's suggestion, we have repeated the co-IP experiments (Fig. IX, below for convenience), including the results of the interaction between CTR9 (1-249) and PAF1  . We have included the newly acquired Fig. IX as the revised Fig. 2f and modified the manuscript accordingly. 3. The authors attempt to define a new mode of interaction among TPR motifs. TPR motifs are widespread in nature and present a common fold. Further evidence would be needed to conclude that a novel TPR-SLIM (small linear motif) interaction has been found. This point seems over-interpreted. Comprehensive bioinformatics analysis and a figure showing a comparison of structural results (and different type of folds) would be necessary to raise this conclusion. Otherwise, the authors are encouraged to moderate (or even omit) this statement in the text.
We fully agree with the reviewer on this comment and have stated the phrase in section of abstract or discussion to "Here we report two structures of each of the human and yeast Paf1/Ctr9 subcomplexes, which assemble into heterodimers with very similar conformations, revealing a previously uncharacterized interface between the tetratricopeptide repeat (TPR) module in Ctr9 and Paf1." 4. To validate the interactions observed in the structure of hCtr91-249/hPaf157-116 a number of mutagenesis studies are presented. The overall conclusion of this study seems precise. However, Figure S6 and related comments need to be improved for clarity. For example, all mutants (except hCtr91-249 (R98W)/hPaf157-116, Figure S6C, blue lane) seem to aggregate. This is a typical behaviour of disrupted complexes that occurs when one or more of the interactor partners are missing. In Figures S6A and S6A2, Ctr9 seems to appear in the void volume (first peak of SEC, S6A) and Paf1 shows up later (second peak of SEC, S6A). In the same figure, it also seems like Ctr9 is also present in this second peak ( Figure S6A2, lane 35) and shows the same intensity as Paf1. This does not correlate with the absorbance of the chromatogram peaks. In addition, in S6A2 there are 2 bands of similar size (14.4 kDa) and it is unclear which one corresponds to hPaf157-116(5M). Could the authors elaborate further on the interpretation of the chromatograms and corresponding gels, and specifically on Figures  S6A and S6A2? Could it be that mutated complexes are still observed in the second peak of the chromatograms (~fraction 35) but with disrupted stoichiometry? See comments on how to improve Figure S6A below.
We fully agree the reviewer on these comments and appreciate the reviewer pointing out some very important concern out. During revision, we have purified CTR9 (1-249) and CTR9  /PAF1 (57-116) (5M). Size-exclusion chromatography and SDS-PAGE analysis showed that the purified CTR9 (1-249) is aggregated in the absence of PAF1  binding (dotted line and blue line in Fig. X, below). Additionally, we found that yeast Ctr9 (1-313) forms inclusion bodies when expressing in E. coli (data now shown). These data indicate that both human CTR9/PAF1 and yeast Ctr9/Paf1 subcomplexes formation are essential for the folding of CTR9 and Ctr9 (at least for CTR9  and Ctr9 (1-313) ), respectively. We have included  5. Lines 85-86 "elucidated the mechanism responsible for human diseases caused by mutations of Paf1 and Ctr9" are an overstatement of the conclusions. The authors have demonstrated that the assembly of PAF1C might have an impact on specific cancer diseases, but further investigation is needed to elucidate the mechanism responsible for the human disease.
We agree the reviewer for this point. We have removed this statement and modified the manuscript accordingly.
Suggestions to improve figures. Figure 1A.
• Summarizing disease-associated mutations on this figure is not constructive. A table showing a summary of mutations is recommended. For simplicity, information about mutations should be omitted in the figure. This way, this figure would compare easily to Figure 3A. Instead, the schematic representation should show all domains present in hCtr9 and hPaf1 (full domain representation) for clarity. Schematic representation of Figure 3A should be modified accordingly.
We fully agree with the reviewer on this point. According to the reviewer's suggestion, we have omitted the mutations information in the original Fig. 1A and shown all domains in CTR9 and PAF1 in Figure XI below. Besides, the full lists of the disease-associated mutations of CTR9 and PAF1 were summarized in Table I and Table II below, respectively. We have included the Figure XI as revised Fig. 1a and the Table I and Table II as the  revised Supplementary Table 2 and Table 3, respectively.

Mutation type
Mutation ( In-frame deletion R962delR Various disease-associated mutations in CTR9 were listed, based on the data extracted from the COSMIC database (http://cancer.sanger.ac.uk/cosmic). A total of 38 missense mutations were located in 38 residues of CTR9  . These 38 mutations were divided into three categories. The three categories were named as interface, folding, and others, and amino acid substitutions in each category were colored in red, orange, and green, respectively. Only mutations used in this study were shown in Fig. 2g.
Various disease-associated mutations in PAF1 were listed, based on the data extracted from the COSMIC database . A total of 21 missense mutations were located in 15 residues of PAF1  . These 21 mutations were divided into three categories. The three categories were named as interface, folding, and others, and amino acid substitutions in each category were colored in red, orange, and green, respectively. Only mutations used in this study were shown in Fig. 2g. According to the reviewer's suggestion, helices in protein structure were shown in cylinders and the color codes CTR9  in cyan, and PAF1  in magenta or Ctr9  in orange and Paf1  in green were added in the revised Fig. 1b and Fig. 2a, or Fig. 3b, respectively. Thanks for suggestion. Following the reviewer's suggestion, in the revised Fig. 2e-d and  Fig. 3e-g, we have used LigPlot to show the residues involved in the interactions between human CTR9 and PAF1 (Fig. XII, below) heterodimer or yeast Ctr9 and Paf1 heterodimer (Fig. XIII, below), respectively. Alternatively, in the revised Supplementary Fig. 5a-f, we have shown the interaction details between CTR9 and PAF1 or Ctr9 and Paf1 with cartoon representation in stereo view.  • Figure 2E. The name of the proteins used to depict secondary structures on top (hPaf1) and bottom (yPaf1) of alignment should be indicated for clarity. Have these secondary structures been predicted with any specific program? The same applies to Figure S2 (hCtr9 and yCtr9).
We have added the protein names at the top and bottom of each alignment shown in Fig. 2e and Supplementary Fig. 2. All sequence alignments shown in this manuscript are based the crystal structures of CTR9 (1-249) -PAF1  and Ctr9 (1-313) -Paf1  , which are determined in this study.
• Figures 2F and 2H. Labels of specific band proteins are not shown in Figures 2F and 2H. Also, is hPaf157-116 tagged with GFP in N or C terminus? (N-terminal residue in Figure 2F and C-terminal one in Figure 2H).
Thanks for pointing out the mistakes. Myc was tagged to the N-terminal of CTR9  and GFP was tagged to the C-terminal of PAF1 (57-116) WT or mutant. For clarify, we have stated the position of each tag in all constructs used in Co-IP experiments.
• Figure 2G. To make it easier understand what type or residues are included in categories 1, 2, 3 (both mentioned in the text and in this figure), these could be renamed as 'interface', 'folding' and 'others', respectively. This figure is difficult to interpret and it is not very informative (again a comprehensive table with mutations would help if added). To make this figure more useful, the authors could focus on showing just the mutations used in this work (categories 1 and 3).
Thanks for suggestion. Following the reviewer's suggestion, we have revised Fig. 2g (Fig.  XIV, below) and modified the manuscript accordingly.

Fig. XIV.
Disease-associated mutations in the CTR9 (1-129) /PAF1  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 Table 2 and 3 (Table I and II, up), respectively. Figure 5E. Label of IB is missing.
We have improved this unclear presentation in the revised Fig. 6e. Figure S1. • Contrast of S1B needs to be improved. • Color code for chromatograms is not shown in S1E.
In the revised Supplementary Fig. 1, the original SDS-PAGE in (b) was replaced by a new one and the color codes in (e) were added. Figure S3. This figure can be used to highlight the differences in the N-terminal domain of the human and yeast proteins, which will complement Figure 3H. This represents an important finding in this paper, i.e. this N-terminal region is essential to maintain the interaction between yCtr9 and yPaf1.
Thanks for reviewer's suggestion. Structural-based sequence alignment of the N-terminal tail of Ctr9 in various species indicated that yeast Ctr9 has a longer N-terminal tail than other species (e.g., human CTR9) (Fig. XVa, below). The Ctr9 (1-313) -Paf1 (34-103) structure showed that residues Y10, P11, M13, E14, and W15 in the longer N-terminal tail of yeast Ctr9 are directly involved in binding to Paf1 (revised Fig. 3f and 3g). Fig. XVb clearly showed that the [Ctr9(N16)] mutant, in which the most N-terminal 16 amino acids of Ctr9 were deleted, could not form complex with Paf1 thoroughly, indicating this longer N-terminal tail of Ctr9 is essential to maintain the interaction between Ctr9 and Paf1. We have included Fig. XV as the revised Fig.4 and modified the manuscript accordingly. In this alignment, the secondary structures of human CTR9 (1-38) and yeast Ctr9  are shown at the top and bottom, respectively, and conserved residues are shaded in red. The amino acids 1-16 of yeast Ctr9, which were deleted in the GFP-Ctr9(N16) 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(N16Δ) 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. Figure S4C. To appreciate the differences between human and yeast Ctr9/Paf1, the use of only two colours would simplify the representation.
Following the reviewer's suggestion, human CTR9/PAF1 is colored in cyan and yeast Ctr9/Paf1 is colored in orange (revised Supplementary Fig. 4c) Figure S5. • S5A and S5B. It is not possible to appreciate the stick representation of residues very well. A zoom highlighting these regions would help the reader. Labels in green and blue are not easy to read. A possibility could be to use lighter colours.
We agree the reviewer on this point. We have removed the original S5A and S5B. In the revised Supplementary Fig. 5a-f, we have shown the interaction details between CTR9 and PAF1 or between Ctr9 and Paf1 with cartoon representation in stereo view. Following the reviewer's suggestion, in the revised Fig. 2e-d and Fig. 3e-g, we have used LigPlot to show the residues involved in the interactions between human CTR9 and PAF1 or yeast Ctr9 and Paf1 heterodimer, respectively.
• Figure S5E. Not mention in the text.
The revised Supplementary Fig. 7g is mentioned in the phrase of "Our structural analyses show that both ligands human PAF1 and yeast Paf1 formed a similar hook-fold" . Thanks for reviewer's suggestion. In the revised Supplementary Fig .6, fractions are labeled in each gel filtration profiles; those fractions shown in correspondent gels are indicated by a two-way arrow; and represent fraction number are wrote in one lane.
Minor points.
-Several parts of the manuscript need to be clarified so it is more suitable for a general audience. This will aid the reader understand better the text. For example, a general overview of what it is known about Paf1, Ctr9 and PAF1C from the structural point of view is missing.
Following the reviewer's suggestions. We have included the structural information about Paf1 complex in the revised introduction. The structures of Ras-like domain of Cdc73 (Refs. #38 and #39 in this manuscript) and the Plus3 domain of RTF1 (Refs. #40, #41, and #42 in this manuscript) 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-jaw tumor mutants (Ref. #43 in this manuscript). The structure of histone modification domain (HMD) of Rtf1 was reported and the HMD was shown to stimulate H2B ubiquitylation through interaction with Rad6 (Ref. #44 in this manuscript). However, there is no atomic structure information has been reported about Ctr9, which contains multi-motifs for protein-protein interaction (Figs. 1a and 3a).
-'Crystal structure of the hCtr9/hPaf1 heterodimer' section. It is unclear how the TEV constructs have been made. Is the TEV site cleaved at all? Specify in line 117 where the TEV-cleavable segment has been included. Explain how the artificial linker is helping crystallization.
Thanks for reviewer for pointing these concern out. We have fused CTR9  to the Nterminus of PAF1 (57-116) with a TEV-cleavable segment to make a single-chain fusion protein of the CTR9 (1-249) -PAF1  complex. DNA fragments were amplified by PCR and linked with a tobacco etch virus (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. Similar strategy was applied to make single-chain fusion protein of Ctr9 (1-313) -Paf1  complex. During crystallization, we did not add any protease to proteins sample and thus SDS-PAGE analysis of dissolved crystals demonstrated that the molecules were intact (revised Supplementary Fig. 3b and 3d). It is possible that the artificial linker may stabilize the N-terminus of PAF1  or Paf1  and thus helping complex proteins crystallization. We have modified the manuscript accordingly.
-Lane 267. Elaborate further on the interaction between the N-terminal tail of yCtr9 and yPaf1 and how this is different to the human system.
Fowling the reviewer's suggestion, we have included a newly Fig. 4a, indicating that yeast Ctr9 has a longer N-terminal tail than other species (e.g., human CTR9) (Fig. XVa, up). The Ctr9 (1-313) -Paf1 (34-103) structure showed that residues Y10, P11, M13, E14, and W15 in the longer N-terminal tail of yeast Ctr9 are directly involved in binding to Paf1 (revised Fig. 3f and 3g). Fig. XVb clearly showed that the [Ctr9(N16)] mutant, in which the most Nterminal 16 amino acids of Ctr9 were deleted, could not form complex with Paf1 thoroughly, indicating this longer N-terminal tail of Ctr9 is essential to maintain the interaction between Ctr9 and Paf1. We have modified the manuscript accordingly.
-Lane 268. Specify that yL83 is tested because it is only present in yeast.
Following the reviewer's suggestion, we have modified the manuscript accordingly.
-Different terms are used for co-immunoprecipitation (co-IP). For example in lines 290 and 294 it is named as coprecipitation. Check consistency for this term.

Fig. XVI.
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 (1-313) -Paf1  heterodimer or the interaction between Paf1 and Leo1 obtained from the IP results and previous study (ref. 45,46 ). The fine lines represent the interaction obtained from the IP results. The dotted lines represent interactions that need to be further studied.
-D95K mutation in yPaf1 abolished the interaction between yCtr9 and yPaf1 completely. It would be interesting to test if D104K in hPaf1 had the same effect.