TRANSCRIPTION

Structure of the complete elongationcomplex of RNA polymerase II withbasal factors

Haruhiko Ehara,1 Takeshi Yokoyama,1 Hideki Shigematsu,1 Shigeyuki Yokoyama,2

Mikako Shirouzu,1 Shun-ichi Sekine1*

In the early stage of transcription, eukaryotic RNA polymerase II (Pol II) exchangesinitiation factors with elongation factors to form an elongation complex for processivetranscription. Here we report the structure of the Pol II elongation complex bound with thebasal elongation factors Spt4/5, Elf1, and TFIIS. Spt4/5 (the Spt4/Spt5 complex) and Elf1modify a wide area of the Pol II surface. Elf1 bridges the Pol II central cleft, completing a“DNA entry tunnel” for downstream DNA. Spt4 and the Spt5 NGN and KOW1 domainsencircle the upstream DNA, constituting a “DNA exit tunnel.” The Spt5 KOW4 and KOW5domains augment the “RNA exit tunnel,” directing the exiting nascent RNA. Thus, theelongation complex establishes a completely different transcription and regulationplatform from that of the initiation complexes.

Transcription is accomplished by DNA-dependent RNApolymerase (RNAP) througha multistep process consisting of initiation,elongation, and termination. In the initia-tion of eukaryoticmRNA transcription, RNA

polymerase II (Pol II) is associated with generaltranscription factors (GTFs), forming a huge ini-

tiation complex (IC) on the DNA (1–3). After thesynthesis of a certain length of RNA, the GTFsare replaced with elongation factors (EFs). Thus,the IC isomerizes into a processive elongationcomplex (EC), which serves as a platform for reg-ulation and various transcription-coupled events,such asmRNAprocessing, chromatin remodeling,

andDNA repair (4–7). Although recent advances incryo–electronmicroscopy (cryo-EM) have revealedseveral IC structures (1–3), the architecture of theEC still remained elusive. Here we elucidated thePol II EC structure, including the conserved basalEFs Spt4/5, Elf1, andTFIIS, by using an integratedapproach combining x-ray crystallography andcryo-EM.Spt4/5, a heterodimeric complex of Spt4 and

Spt5 (NusG in bacteria), is implicated in pro-cessive transcription elongation (8, 9), promoter-proximal pausing in higher eukaryotes (as DRBsensitivity-inducing factor) (8, 10), and manytranscription-coupled events (4–7). Spt5 is amulti-domain protein composed of the conservedNGN(NusGN-terminal) domain,which binds Spt4, anddifferent numbers of KOW (Kyprides-Onzonis-Woese) domains, which are connected by flexiblelinkers and the C-terminal repeat region (CTR)(Fig. 1A) (11–13).Multiple KOWdomains specific toeukaryotic Spt5 are crucial for the establishmentof the EC and Spt4/5 functions (14, 15), but theirlocations on Pol II and functions in transcriptionhave remained enigmatic. To assess which KOWdomain is essential for Pol II binding and tran-scription, we created a series of Spt4/5 variantsand investigated their transcription-stimulationfunctions for Pol II from the yeastKomagataellapastoris (16) (Fig. 1, A and B). Surprisingly, only

RESEARCH

Ehara et al., Science 357, 921–924 (2017) 1 September 2017 1 of 4

1RIKEN Center for Life Science Technologies, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan. 2RIKENStructural Biology Laboratory, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan.*Corresponding author. Email: shunichi.sekine@riken.jp

Fig. 1. Crystal structure of Pol II EC with Spt5 KOW5 and Elf1. (A) TheSpt4/5 and Elf1 constructs used in this study (numbers represent aminoacid residue positions). The regions studied in the crystallographic andcryo-EM analyses are indicated by the red and green boxes, respectively.(B) Polyacrylamide gel electrophoresis analysis of in vitro transcriptionstimulation by Spt4/5 variants. (C) Overall structure of the Pol II EC–

KOW5-Elf1 complex. (D) Close-up views of the Spt5 KOW5–binding site intwo different orientations. (E) The Elf1-binding site and the DNA entrytunnel, shown in two different orientations. In the right panel, the N-terminal basic tail of Elf1 was presumed and is depicted as a dashed line.(F) The amino acid sequence of the Elf1 N-terminal tail, in which the basicresidues are highlighted. b1 represents the first b strand of Elf1.

on April 27, 2020

http://science.sciencem

ag.org/D

ownloaded from

KOW5 is necessary and sufficient for the transcrip-tion stimulation,whereas the otherKOWdomains,the NGN domain, and Spt4 were dispensable.These data underscore the importance of KOW5in the Spt4/5 function as an EF.Elongation factor 1 (Elf1 in yeast; ELOF1 in

human) is a small protein conserved in eukaryotesand several archaeal classes, including Crenarch-aeota (fig. S1) (17). Elf1 displays synthetic lethalitywith several EFs, including Spt4, Spt5, and TFIISin yeast (18). Chromatin immunoprecipitationanalyses revealed that Elf1 accompanies tran-scribing Pol II in vivo (18, 19). As Elf1 slightlydelays transcription elongation by K. pastorisPol II in vitro, it seems to directly interact withPol II and function in transcription regulation(fig. S2).To understand how the Spt5 KOW5 domain

and Elf1 associate with Pol II, we solved thecocrystal structure of the K. pastoris Pol II ECbound with KOW5 and Elf1 at 3.0 Å resolution

(Fig. 1C, fig. S3, and table S1). KOW5 binds in acomplementary pocket formed at the junctionof five Pol II subunits: Rpb1, Rpb2, Rpb3, Rpb11,and Rpb12 (Fig. 1D), which interact extensivelythrough electrostatic and polar interactions be-tweenwell-conserved amino acid residues (fig. S4).Notably, KOW5 bridges the Rpb1 dock domainand the Rpb2 wall domain of Pol II, becomingan integral part of the “RNA exit tunnel,” whichis described later in the text. The location of KOW5constrains the relative orientations of the shelfand coremodules (20) of Pol II. Thus, KOW5 alsoseems to adjust the Pol II conformation (21) so itis suitable for processive elongation.Elf1 is located between the Rpb2 lobe domain

and the Rpb1 clamp-head domain of Pol II (Fig.1E). While the a helix (the Pol II–binding helix)of Elf1 interacts with the Arg-rich face at the tipof the Rpb2 lobe (fig. S5), the opposite side of Elf1reaches the Rpb1 clamp head. As a consequence,Elf1 fills the gap between the core and clamp

modules over the central cleft of Pol II and thuscompletes a closed “DNA entry tunnel” for thepassage of the downstreamDNA duplex. Althoughthe N-terminal basic tail is disordered, it shouldinteract with the DNA held in the closed tunnel(Fig. 1, E and F). Even though Elf1 delays tran-scription in vitro, itmay stabilize theECbyprevent-ing DNA dissociation. Consistently, the effect ofElf1 was compromised when the N-terminal tailwas deleted or the Pol II–interacting residues weremutated (fig. S2).Althoughwe identified the binding site of Spt5

KOW5 on Pol II, the positions and nucleic acidinteractions of the other KOW domains, theNGN domain, and Spt4 were still unclear. Toaddress these problems, we performed cryo-EMsingle-particle analyses of the Pol II EC boundwith Spt4/5, including an Spt5 variant (residues217 to 815, from NGN to KOW5) and an inactivevariant of TFIIS (table S2). The three-dimensionalreconstruction with 682,749 particles yielded the

Ehara et al., Science 357, 921–924 (2017) 1 September 2017 2 of 4

Fig. 2. Cryo-EM structure of Pol II EC with Spt4/5 and TFIIS.Overall structure of the complex showing (A) the “DNA-binding”face and (B) the “RNA-exit” face. (C) Interactions of Spt4 and Spt5NGN-KOW1 with Pol II and DNA. The cryo-EM map (before sharpening)corresponding to Spt4, Spt5 NGN-KOW1, and the upstream DNA isoverlaid. (D) Top view of the DNA exit tunnel. (E) Side view of theDNA exit tunnel. (F) The same view as in (E) with the surface

electrostatic potentials of Spt4 and Spt5 NGN-KOW1. (G) Close-upview of the RNA exit site. The cryo-EM map (before sharpening)corresponding to the exiting RNA and Spt5 KOW4-KOW5 is overlaid. TheKOW4-KOW5 linker and a possible path of the exiting RNA are depictedby a green line and a red dashed arrow, respectively. (H) Surfaceelectrostatic potentials of the Spt5 KOW4-KOW5 domains, the Rpb1dock, and the Rpb4/7 stalk.

RESEARCH | REPORTon A

pril 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

Pol II EC structure at an overall resolution of3.8 Å (fig. S6). Structural heterogeneities wereobserved around the upstream DNA, the Pol IIRpb4/7 stalk, and the secondary channel. Fo-cused classifications for these regions success-fully resolved the upstream DNA-bound Spt4and Spt5 NGN-KOW1, the stalk-bound Spt5KOW4, and the secondary channel–bound TFIIS,respectively (figs. S7 to S11).In the cryo-EM structure, the Spt5 NGN domain

is docked to the tip of the Rpb1 clamp coiled coiland occupies the space between the clamp (Rpb1coiled coil) and core (Rpb2 protrusion and lobe)modules (Fig. 2, A and B). The closed clamp co-incides with the fluorescence resonance energytransfer study on an archaeal RNAP (22), and theNGN position is generally consistent with thatseen in the partial x-ray and 13 Å cryo-EM struc-tures of the archaeal RNAP and the 26 Å negative-stain EM structure of mammalian Pol II (23–25).

KOW1 is bound to the “zipper” loop of the Rpb1clamp (Fig. 2, C and D). In addition, the eukaryote-specific insertion of KOW1 contacts the tip of theRpb2 wall. Thus, KOW1 bridges the clamp andcore modules of Pol II on the other side of theNGN domain. Spt4 is tightly associated with theNGN and resides between KOW1 and the Rpb2protrusion.As a consequence, theSpt5NGN-KOW1domains

and Spt4, together with the Rpb2 protrusion andwall, establish a “DNA exit tunnel” that coversapproximately one helical pitch of the upstreamDNA duplex (Fig. 2, C and E). The conservedArg232, Lys235, Lys237, Arg241, and Lys269 residuesof the NGN, together with Lys416, Arg419, Arg423,Arg427, Arg433, Lys442, and Lys463 of the Rpb2protrusion, form a positively charged surface,which contacts the proximal five base pairs ofthe upstream DNA duplex (positions –14 to –10)(Fig. 2F and fig. S12). In addition, three loops of

the NGN (residues 230 to 235, 265 to 270, and286 to 296) contact the single-stranded part ofthe nontemplate DNA (positions –5 to –9) in thetranscription bubble, consistent with the previ-ous cross-linking study (14). KOW1 also providesa surface rich in basic and polar residues (Lys329,Asn330, Lys334, Lys386, Lys428, Arg431, Gln433, andAsn434), throughwhich it interactswith the distalpart of the upstream DNA duplex (positions –17to –12). The b1-b2 loop of KOW1 (residues 328 to335) fits in theminor groove of the DNA (fig. S13).Spt4 and the eukaryote-specific insertion of

KOW1 loosely surround the distal part of theDNA. Although they are farther apart from theDNA, they contain several basic residues thatcould potentially participate in DNA binding.Thus, the DNA exit tunnel assumes a funnel-likeshape, which allows a certain degree of DNAbend-ing in its distal part while restricting the DNAorientation in the proximal part. The focusedclassification for the upstream DNA depicted sev-eral KOW1-lacking classes with different DNAandNGN orientations (fig. S14). This emphasizesthe importance of KOW1 for defining the appro-priate DNA orientation and explains why the dele-tion of KOW1 is lethal in yeast (14).Spt5 KOW4 is bound to the oligonucleotide/

oligosaccharide-binding (OB) fold domain of thestalk subunit Rpb7 (Fig. 2B and fig. S15). The lo-cation of KOW4 is in good agreement with theprevious cross-linking study (7). Thus, Spt5KOW1,KOW4, KOW5, and Rpb7 are distributed aroundthe rim of the RNA exit tunnel, expanding it intoa large funnel-like structure. Although the tandemlypacked KOW2 and KOW3 domains (12) were notresolved, they should reside between KOW1 andKOW4, possibly on the Rpb1 clamp according tothe cross-linking study (7), and could participatein the expanded RNA exit tunnel.The density for the exiting RNA is observed

along a channel composed of the Spt5 KOW5, theRpb1 dock, and the Rpb2 wall (Fig. 2G). This viewis consistent with the observations that KOW5 iscross-linked to nascent RNA (15) and importantfor transcription processivity (14). AsKOW5blocksone of the twopreviously proposedRNAexit paths(path 2) (20, 26), the RNA should be directedthrough the remaining path 1 toward the Pol IIstalk. A weak density for the linker connectingKOW4 and KOW5 was observed, which also par-ticipates in the path. A long stretch of nascentRNA is required for the stable binding of Spt4/5to the Pol II EC (27), and the RNA interacts withthe stalk subunit Rpb7 (26). KOW4 cooperateswith Rpb7 to form a basic patch (Fig. 2H), whichcould serve as the RNA-binding site. Thus, theSpt5 KOWs augment the RNA exit tunnel, likelysupporting the efficient exit of the nascent RNA.The combination of the present x-ray and cryo-

EM structures provides a reliable compositemodelof the Pol II EC, including the basal EFs Spt4/5,Elf1, and TFIIS (Fig. 3 andmovie S1). The EFs aredistributed on a wide area of the Pol II surfaceand constitute or expand the DNA entry tunnel(Elf1), the DNA exit tunnel (Spt4 and Spt5 NGN-KOW1), and the RNA exit tunnel (Spt5 KOW1 to5). Spt5 NGN and Elf1 are adjacent to each other

Ehara et al., Science 357, 921–924 (2017) 1 September 2017 3 of 4

Fig. 3. The architecture of the Pol II EC. (A) The DNA and RNA binding sites in the EC (the EFsare omitted). (B) The same as (A) but with bound Spt4/5, Elf1, and TFIIS (the composite modelshowing the DNA entry and exit tunnels). (C) The view of the RNA exit tunnel (the EFs are omitted).(D) The same as (C) but with the EFs (the expanded RNA exit tunnel). The KOW4-KOW5 linker, Pol IICTD, and Spt5 CTR are depicted by green, brown, and violet lines, respectively. (E) Schematicdiagram summarizing the interactions between nucleic acids and EFs.

RESEARCH | REPORTon A

pril 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

at the inner corner of the 90°-bent DNA. Spt4/5and Elf1 achieve almost symmetrical coverage ofthe upstream and downstream parts of the DNA,encompassing it from positions –15 to +16 (Fig.3E). The established DNA entry and exit tunnelsenable proper DNA unwinding and rewinding inthe defined size of the transcription bubble. Spt5KOW1, possibly with the unresolved KOW2 andKOW3, physically separates theDNAandRNAexitfaces, preventing potential DNA-RNA interactionssuchasR-loop formation.These features extensivelyrepresent the architecture required for proces-sive transcription, while counteracting pausing,backtracking, arrest, and premature termination.The KOW5 location suggests that the CTR, theintrinsically unstructured extension of Spt5, pro-trudes from the rim of the RNA exit tunnel (Fig.3D). This is quite reasonable, as the CTR, as wellas the Pol II C-terminal domain (CTD), recruitsvarious factors, including mRNA-processing ma-chineries (4–6, 28, 29).A structural comparison reveals thatmany com-

ponents of the EC are mutually exclusive withthose of ICs (1–3) (fig. S16). Spt4, Spt5 NGN, andElf1, which seal the Pol II cleft, overlap with theDNA duplex in the preinitiation complex. More-over, Spt4 and Spt5 KOW4 in the EC overlapwiththe RAP30 subunit of TFIIF and the a subunit ofTFIIE, respectively, in ICs. These observations indi-cate that the EC is only achieved after the isom-erization from IC to EC and the exchange ofGTFs with EFs. Consistently, the exchange ofarchaeal TF(II)E and Spt4/5 was reported (13, 30).Collectively, the EC establishes a different ar-chitecture from those of ICs, for processivetranscription, regulation, andmany transcription-coupled events.

REFERENCES AND NOTES

1. Y. He et al., Nature 533, 359–365 (2016).2. C. Plaschka et al., Nature 533, 353–358 (2016).3. K. Murakami et al., Proc. Natl. Acad. Sci. U.S.A. 112,

13543–13548 (2015).4. K. Zhou, W. H. Kuo, J. Fillingham, J. F. Greenblatt, Proc. Natl.

Acad. Sci. U.S.A. 106, 6956–6961 (2009).5. A. Mayer et al., Mol. Cell. Biol. 32, 1321–1331 (2012).6. S. Schneider, Y. Pei, S. Shuman, B. Schwer, Mol. Cell. Biol. 30,

2353–2364 (2010).7. W. Li, C. Giles, S. Li, Nucleic Acids Res. 42, 7069–7083

(2014).8. T. Wada et al., Genes Dev. 12, 343–356 (1998).9. A. Hirtreiter et al., Nucleic Acids Res. 38, 4040–4051

(2010).10. J. Li, D. S. Gilmour, Curr. Opin. Genet. Dev. 21, 231–235

(2011).11. C. P. Ponting, Nucleic Acids Res. 30, 3643–3652

(2002).12. P. A. Meyer et al., Mol. Cell. Biol. 35, 3354–3369 (2015).13. F. Werner, J. Mol. Biol. 417, 13–27 (2012).14. J. B. Crickard, J. Fu, J. C. Reese, J. Biol. Chem. 291,

9853–9870 (2016).15. Y. Qiu, D. S. Gilmour, J. Biol. Chem. 292, 5555–5570

(2017).16. H. Ehara, T. Umehara, S. I. Sekine, S. Yokoyama, Biochem.

Biophys. Res. Commun. 487, 230–235 (2017).17. J. P. Daniels, S. Kelly, B. Wickstead, K. Gull, Biol. Direct 4, 24

(2009).18. D. Prather, N. J. Krogan, A. Emili, J. F. Greenblatt, F. Winston,

Mol. Cell. Biol. 25, 10122–10135 (2005).19. A. Mayer et al., Nat. Struct. Mol. Biol. 17, 1272–1278

(2010).20. P. Cramer, D. A. Bushnell, R. D. Kornberg, Science 292,

1863–1876 (2001).21. S. Sekine, Y. Murayama, V. Svetlov, E. Nudler, S. Yokoyama,

Mol. Cell 57, 408–421 (2015).22. S. Schulz et al., Proc. Natl. Acad. Sci. U.S.A. 113, E1816–E1825

(2016).23. B. J. Klein et al., Proc. Natl. Acad. Sci. U.S.A. 108, 546–550

(2011).24. C. Bernecky, F. Herzog, W. Baumeister, J. M. Plitzko, P. Cramer,

Nature 529, 551–554 (2016).25. F. W. Martinez-Rucobo, S. Sainsbury, A. C. Cheung, P. Cramer,

EMBO J. 30, 1302–1310 (2011).

26. A. Újvári, D. S. Luse, Nat. Struct. Mol. Biol. 13, 49–54(2006).

27. A. Missra, D. S. Gilmour, Proc. Natl. Acad. Sci. U.S.A. 107,11301–11306 (2010).

28. T. Yamada et al., Mol. Cell 21, 227–237 (2006).29. Y. Liu et al., Mol. Cell. Biol. 29, 4852–4863 (2009).30. D. Grohmann et al., Mol. Cell 43, 263–274 (2011).

ACKNOWLEDGMENTS

We thank Y. Tomabechi, T. Uchikubo-Kamo, and T. Osanaifor technical assistance. The synchrotron radiation experimentswere performed at BL41XU of SPring-8 with the approvalof the Japan Synchrotron Radiation Research Institute(proposal nos. 2014B1265, 2015B2040, 2016A2526, and2016B2526) and at NE3A of the Photon Factory with theapproval of the Photon Factory Program Advisory Committee(proposal no. 2015G520). We thank H. Okumura andK. Hasegawa for assistance with data collection at SPring-8.This work was supported in part by the Japan Society for thePromotion of Science KAKENHI grant numbers JP15H04344(to S.S.) and JP15H01656 (to H.S.), the Platform Projectfor Supporting Drug Discovery and Life Science Researchfunded by the Japan Agency for Medical Research andDevelopment, and the RIKEN Pioneering Project, DynamicStructural Biology. The atomic coordinates and structurefactors for the EC bound with Elf1 and Spt5 KOW5have been deposited in the Protein Data Bank(PDB ID 5XOG). The cryo-EM density map and theatomic coordinates for the EC bound with Spt4/5 andTFIIS were deposited in the Electron Microscopy (EM)Data Bank (accession code EMD-6747) and the ProteinData Bank (PDB ID 5XON), respectively.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/357/6354/921/suppl/DC1Materials and MethodsFigs. S1 to S16Tables S1 and S2References (31–45)Movie S1

30 May 2017; accepted 24 July 2017Published online 3 August 201710.1126/science.aan8552

Ehara et al., Science 357, 921–924 (2017) 1 September 2017 4 of 4

RESEARCH | REPORTon A

pril 27, 2020

http://science.sciencemag.org/

Dow

nloaded from

Structure of the complete elongation complex of RNA polymerase II with basal factorsHaruhiko Ehara, Takeshi Yokoyama, Hideki Shigematsu, Shigeyuki Yokoyama, Mikako Shirouzu and Shun-ichi Sekine

originally published online August 3, 2017DOI: 10.1126/science.aan8552 (6354), 921-924.357Science 

, this issue p. 921; see also p. 871Scienceor rewinding. The Pol II elongation complex thus adopts a stable architecture suitable for processive transcription.establish an RNA exit path and DNA entry or exit tunnels, which facilitate nascent transcript transfer and DNA unwinding (see the Perspective by Fouqueau and Werner). Multiple elongation factors are distributed on a wide surface of Pol II andthe high-resolution structure of the elongation complex by means of x-ray crystallography and cryo-electron microscopy

determinedet al.with several other factors to form an elongation complex that promotes transcription elongation. Ehara Eukaryotic transcription of mRNA is a multistep process mediated by RNA polymerase II (Pol II). Pol II combines

Transcription machinery remains steadfast

ARTICLE TOOLS http://science.sciencemag.org/content/357/6354/921

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2017/08/03/science.aan8552.DC1

CONTENTRELATED http://science.sciencemag.org/content/sci/357/6354/871.full

REFERENCES

http://science.sciencemag.org/content/357/6354/921#BIBLThis article cites 45 articles, 14 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Copyright © 2017, American Association for the Advancement of Science

on April 27, 2020

http://science.sciencem

ag.org/D

ownloaded from