TRANSCRIPTION

Structures of the RNA polymerase-s54

reveal new and conservedregulatory strategiesYun Yang,1,4* Vidya C. Darbari,1,2* Nan Zhang,3 Duo Lu,1,2† Robert Glyde,1,2

Yi-Ping Wang,4 Jared T. Winkelman,5 Richard L. Gourse,5 Katsuhiko S. Murakami,6

Martin Buck,3 Xiaodong Zhang1,2‡

Transcription by RNA polymerase (RNAP) in bacteria requires specific promoterrecognition by s factors. The major variant s factor (s54) initially forms a transcriptionallysilent complex requiring specialized adenosine triphosphate–dependent activators forinitiation. Our crystal structure of the 450-kilodalton RNAP-s54 holoenzyme at 3.8angstroms reveals molecular details of s54 and its interactions with RNAP. The structureexplains how s54 targets different regions in RNAP to exert its inhibitory function.Although s54 and the major s factor, s70, have similar functional domains and contactsimilar regions of RNAP, unanticipated differences are observed in their domainarrangement and interactions with RNAP, explaining their distinct properties.Furthermore, we observe evolutionarily conserved regulatory hotspots in RNAPs that canbe targeted by a diverse range of mechanisms to fine tune transcription.

Gene transcription is a tightly regulatedevent. A range of factors are required tomaintain RNAP in a signal-responsive,but inhibited, state and to target it to spe-cific genes (1). Bacterial sigma (s) factors,

eukaryotic TFIIB, and other general transcriptionfactors, are the primary promoter recruitmentfactors. The major variant s54 (also called sN) isused in transcribing genes for numerous stressresponses (2). Unlike the major s70 class, whichrecognizes the –35 and –10 promoter DNA ele-ments, the s54 class directs RNAP to promotersites through the –24 and –12 DNA elements andforms a stable closed promoter complex that isunable to spontaneously melt DNA and initiatetranscripts (3). Instead, the initiation processrequires adenosine triphosphate (ATP)–dependentactivator proteins bound to upstream enhancersites to actively remodel the RNAP-s54–DNA com-plex (4).Our crystal structure of RNAP-s54 reveals that

s54 contains four structural domains connectedby long coils and loops that span a large area ofthe RNAP core enzyme, which consists of two aand b, b′, and w subunits (5, 6) (Fig. 1, A to C; fig.

S1, A and B; and table S1). s54 Region I (RI,residues 1–56) forms a hook composed of two ahelices. Region II (RII, residues 57–120) alsocontains two a helices (residues 57–85) in ad-dition to loops that are buried inside theRNAP (Fig. 1, A and C). The RNAP core-bindingdomain (CBD) of s54 extends as a structuralfold to residue 250 and consists of two a-helicalsubdomains. Following on from the CBD, thebackbone extends back to connect to a loopregion; before an extra-long a helix (residues315–353, hereafter called ELH, spanning 50 Å);followed by theHTHdomain (residues 365–385),involved in interaction with the –12 promoterelements (3). The domain containing the RpoNbox (RpoN domain), which is responsible forrecognizing the –24 promoter elements, consistsof a three-helical bundle (residues 415–477) (Fig.1A) (3). The s54 polypeptide chain snakes backand forth through its loop regions embeddedin the RNAP. We carried out cross-linking ex-periments between s54 and RNAP in solutionusing a p-benzoyl-L-phenylalanine–incorporatedRNAP library. The amino acids that are cross-linked to s54 agree well with the RNAP-s54 crys-tal structure (Fig. 1D, yellow, and fig. S1C).We crystallized RNAP-s54 in complex with a

28–base pair (bp) DNA containing the –24 and –12promoter elements. The crystals diffracted to 8 Åresolution, the electron density for DNA was un-ambiguous (Fig. 1E), and the molecular envelopefor the holoenzyme was clear. We thus con-structed a structural model of RNAP-s54–DNA(Fig. 1E and fig. S1D).All s factors contain a major RNAP CBD and

a major DNA binding domain, which recognizeeither the –35 or –24 elements. The CBD of s54

(Fig. 1A) binds toward the upstream face of theRNAP (Fig. 1C) (on the basis of the promoterDNAbinding orientation), which interacts exten-

sively with many functional modules withinthe RNAP, including the b flap (residues 835–937),theC terminusof theb subunit (residues 1267–1320),the b′ zipper/Zn-bindingdomain (residues 35–100),and the b′ dock domain (residues 370–420), aswell as the a-subunit carboxyl-terminal domain(a-CTD) (Fig. 2A). Using FeBABE cleavage as-says, we have previously mapped residue 198of s54 (within the CBD) to a region within the bflap (7).The RpoNdomain of s54 is themost conserved

domain among s54 fromdifferent organisms (fig.S2). This domain extends from the main body ofRNAP and does not contact other parts of s54 orcore enzyme (Fig. 1, A and C). Instead, it interactswith an adjacent RNAP molecule in the crystal,which suggests that its location is flexible in sol-ution. In the RNAP-s54–DNA complex model,the RpoNdomain is indeedmoved relative to theRNAP-s54 structure (fig. S1D). Unlike the flexi-ble RpoN domain, the RIII-HTH domain of s54,which binds to the –12 promoter region, stablyassociates with the RNAP core through RI–RIII(Fig. 1E and fig. S1D) (8–10). Our data thussuggest that –12 binding is a major functionaldeterminant for the s54 holoenzyme promoterrecognition, as well as a stable closed complexformation.The s54 RI plays an inhibitory role and con-

tains contact sites for its cognate activator pro-teins (11, 12). RI interacts with RIII, forming astructural module that lies along the cleft be-tween b and b′ (Figs. 1C and 2B), where tem-plate strand DNA enters the active-site cleft(13). Our structure clearly suggests that theentry of promoter DNA into RNAP is blockedby RI and RIII of s54. The way the RNAP-s54

crystal structure fits into the cryoelectron mi-croscopy map of activator-bound RNAP-s54 holo-enzyme (14) positions the RI helices inside theconnecting density leading to the activatorprotein, which indicates that RI in the RNAP-s54

structure is presented favorably to contact theactivator (fig. S3A).RII penetrates deeply into the DNA binding

channel (Fig. 1 and Fig. 2, C and D). RII can bedivided into three subregions based on theirlocations in the holoenzyme. RII.1 (residues57–85) occupies the downstream DNA bindingcleft of RNAP, just above the bridge helix, andthus plays an inhibitory role (Fig. 1 and Fig. 2, CandD, and fig. S3A, right). RII.2 (residues 86–105)occupies the space of the DNA template strand(Fig. 2, C and D), which indicates that RII.2 hasto relocate to permit template-strand DNA ac-cess into the RNAP active site and for transcrip-tion initiation. RII.3 (residues 105–120) extendsalong the path that is occupied by RNA in thetranscribing complex (Fig. 2, C and D), whichsuggests that it will also need to relocate uponRNA synthesis.Comparisons between s54 and s70 holoenzyme

structures reveal that, overall, they contact similarregions of the RNAP core enzyme (Fig. 3, A and B),in agreement with the idea that specific factorsunrelated by structure and sequence can befunctionally similar (15). However, the relative

882 21 AUGUST 2015 • VOL 349 ISSUE 6250 sciencemag.org SCIENCE

1Centre for Structural Biology, Imperial College London,South Kensington SW7 2AZ, UK. 2Department of Medicine,Imperial College London, South Kensington SW7 2AZ, UK.3Department of Life Sciences, Imperial College London,South Kensington SW7 2AZ, UK. 4State Key Laboratory ofProtein and Plant Gene Research, College of Life Sciences,Peking University, China. 5Department of Bacteriology,University of Wisconsin, Madison, WI 53706, USA.6Department of Biochemistry and Molecular Biology, Centerfor RNA Molecular Biology, Pennsylvania State University,University Park, PA 16802, USA.*These authors contributed equally to this work. †Present address:State Key Laboratory of Bioactive Substance and Function ofNatural Medicine, Institute of Materia Medica, Chinese Academy ofMedical Sciences and Peking Union Medical College, China.‡Corresponding author. E-mail: xiaodong.zhang@imperial.ac.uk

RESEARCH | REPORTSon June 22, 2020

http://science.sciencemag.org/

Dow

nloaded from

location of functional domains is completelydifferent. For example, the s54 CBD is locatedupstream, which blocks the RNA exit tunnel(Fig. 3C), whereas s2, the major CBD of s70,contacts the downstream b′ region (Fig. 3, Aand B). The structural differences suggest thats54 CBD, and hence s54, will have to dissociate

or relocate to allow progression from trans-cription initiation to elongation, whereas s70

can loosely associate with the RNAP coreeven during transcript elongation, in agree-ment with earlier biochemical data (16, 17). s70

region 4 (s4), which recognizes the –35 pro-moter element, occupies the upstream sur-

face blocking the RNA exit channel instead(Fig. 3B).The inhibitory RI of s54 interacts with RIII

across the RNAP cleft and blocks the templatestrand from entering. On the other hand, s1.1of s70, which also has an inhibitory role (18),is located in the downstream DNA channel,

SCIENCE sciencemag.org 21 AUGUST 2015 • VOL 349 ISSUE 6250 883

Fig. 1. Structure of complete RNAP-s54. (A) Structure and schematic diagram of s54. (B) Structure of s70 (PDB 4YG2) with similar functional domains coloredaccordingly. (C) Holoenzyme in different orientations, RNAP core in cylinders. b, cyan; b′, salmon pink; a, gray; w, pale blue; s54 in ribbon and colored as in(A); DS, downstream; US, upstream. (D) Residues of b and b′ subunits that are cross-linked to s54 are mapped onto the structure and colored in yellow.(E) Electron density map and structural model of RNAP-s54–DNA.

RESEARCH | REPORTSon June 22, 2020

http://science.sciencemag.org/

Dow

nloaded from

overlapping with the region occupied by s54

RII.1 (Fig. 3), which suggests that s1.1 and RII.1might play related roles in transcription regu-lation. However, s1.1 connects with s2 by only asingle flexible linker, which allows the s1.1 tomove readily from the RNAP cleft (Fig. 1B, Fig.3B). s54 RII.1 is connected to the RI and CBD. RIinteracts with RIII-ELH, whereas the linker thatconnects RII.1 with the CBD is embedded in theDNA/RNA channel of RNAP (Fig. 2, C and D).Because of its topological restraint, s54 RII.1cannot relocate easily without affecting the con-formation of other parts of s54; RII.1 will thusblock downstream DNA entry.Notable conformational differences exist be-

tween the s70 and s54 holoenzymes in manykey modules within the RNAP when super-posed on the bridge helix, chosen because thecatalytic center must be conserved (fig. S3B).The b′ coiled coil, with the whole b′ subunit,is rotated, which narrows the downstreamDNA channel by as much as 5 Å (fig. S3B).However, the clamp, which consists of b and b′domains forming the walls of the downstreamDNA channel, was shown to adopt multiple con-formations in solution (19). The b′ coiled coilforms helical bundles with s2 of s

70 and helps

884 21 AUGUST 2015 • VOL 349 ISSUE 6250 sciencemag.org SCIENCE

Fig. 2. Functional domains of s54 and interactions with RNAP core. (A) Detailed view of the CBD andRNAP interactions. (B) RI and RIII. (C) DNA-RNA channels, with DNA, orange, and RNA, magenta. (D) Overlaywith RII.1, RII.2, and RII.3.

Fig. 3. Comparisons of s54 and s70 holoenzymes. (A) RNAP-s54

with RNAP displayed as spheres and s54 as surface. (B) RNAP-s70,same view as in (A). (C) Elongation complex (PDB code 2O5J).

RESEARCH | REPORTSon June 22, 2020

http://science.sciencemag.org/

Dow

nloaded from

to stabilize the unpaired nontemplate DNA strandand, hence, the transcription bubble (20) (fig. S4A).The B linker of TFIIB also interacts with the b′coiled coil (fig. S4B). In the s54 holoenzyme struc-ture, the b′ coiled coil does not form a helicalbundle with s54 (fig. S4C). However, it is pos-sible that enroute to transcription initiation,this interaction is established, in agreementwith our biochemical data (fig. S4D).s54 occupies multiple locations along the DNA

and RNA path within the RNAP core. Theselocations are also targeted by other regulatoryfactors and in other RNAPs. For example, RIand RIII sit across the RNAP cleft, blocking theDNA template strand entry (Fig. 4A). At similarlocations, domains 2 and 3 (s2 and s3) of s

70

form a V-shaped wedge that acts as a gate toallow the template strand to enter the activecleft (Fig. 4B) (13, 21). TFIIB also uses two sub-domains, the B core and the B linker, to occupysimilar areas in Pol II (Fig. 4C). This area isthus occupied by s54 to inhibit, while guidingthe template strand delivery by s70 and TFIIB.Furthermore, s2 of s70 contains a number ofaromatic residues that are shown to interactwith the nontemplate strand and therefore fa-cilitate DNA melting and transcription bubbleformation (20). In s54, there are few aromaticresidues in this region (RI and RIII–ELH) thatcould facilitate DNA melting, in agreement withdata showing that although deletion of s54 RIcan bypass the requirement of activator pro-teins, it fails on double-stranded DNA (11). Acti-vator proteinsmay therefore be directly requiredfor transcription bubble opening. The down-stream DNA channel is occupied by both RII.1 ofs54 and s1.1 of s

70 (Fig. 4D). A negatively chargedregion of TFIIF is also located in the downstream

DNA channel of Pol II (22). RII.2 of s54 is locatedin the channel that is occupied by the templatestrand DNA in the transcribing RNAP (Fig. 2D).Both s70 and TFIIB lack equivalent structuralfeatures. This area, however, is occupied in Pol Iby part of the expander (Fig. 4E), a uniqueinsertion in Pol I compared with Pol II andshown to stabilize an expanded active centercleft (23, 24) (Fig. 4E). Similar to RII.3, bothregion 3.2 in s70 (the s finger) and the B readerin TFIIB are shown to overlap with the spacethat would be occupied by RNA (Fig. 4, D andF) (25). Indeed, RII.3 shares sequence homol-ogy with the s finger, especially the highly con-served DDE motif (fig. S2). Acidic residues inthese elements are proposed to facilitate tem-plate DNA loading and RNA separation fromDNA and to guide it toward the exit channel(25–28). It is possible that RII.3 also performssimilar roles.Our structure uncovers a diverse range of

regulatory strategies used by s54 in tightly con-trolling transcription initiation. These includeblocking the template strand from entering theRNAP active cleft by RI and RIII, occupying thedownstream DNA channel by RII.1, and inter-fering with the template strand and synthe-sized RNA by RII.2–RII.3. The position of RIand RIII plays a vital role in these regulatoryfunctions. Activator proteins interact with RIand could relocate RI, RIII–ELH, and RII andrelease the inhibition posed by some of thesestructural elements (fig. S3A) (29, 30). Preciselyhow activators overcome these multiple modesof inhibition, as well as the role of ATP in thisprocess, remains to be determined. Further-more, we show that although s54, s70, TFIIB,TFIIF, and Pol I subunits have no sequence

or structural similarities, they target the sameelements within their respective RNAPs. Ourstructure and comparisons thus provide clearevidence that regulatory hotspots within RNAPsare functionally conserved. Different transcriptionfactors, irrespective of their structures and se-quences, are used to target these hotspots inorder to exert a fine-tuned transcription regu-lation strategy (fig. S5). It will be interesting tosee the extent to which additional multisubunitRNAPs use these regulatory strategies.

REFERENCES AND NOTES

1. R. A. Mooney, S. A. Darst, R. Landick,Mol. Cell20, 335–345 (2005).2. A. Feklístov, B. D. Sharon, S. A. Darst, C. A. Gross, Annu. Rev.

Microbiol. 68, 357–376 (2014).3. M. Buck, M. T. Gallegos, D. J. Studholme, Y. Guo,

J. D. Gralla, J. Bacteriol. 182, 4129–4136 (2000).4. M. Rappas, D. Bose, X. Zhang, Curr. Opin. Struct. Biol. 17,

110–116 (2007).5. P. Cramer, Curr. Opin. Struct. Biol. 12, 89–97 (2002).6. F. Werner, D. Grohmann, Nat. Rev. Microbiol. 9, 85–98 (2011).7. S. R. Wigneshweraraj et al., EMBO J. 23, 4264–4274 (2004).8. Y. Guo, J. D. Gralla, Proc. Natl. Acad. Sci. U.S.A. 95,

11655–11660 (1998).9. Y. Guo, C. M. Lew, J. D. Gralla, Genes Dev. 14, 2242–2255 (2000).10. P. C. Burrows, S. R. Wigneshweraraj, M. Buck, J. Mol. Biol.

375, 43–58 (2008).11. J. T. Wang, A. Syed, J. D. Gralla, Proc. Natl. Acad. Sci. U.S.A.

94, 9538–9543 (1997).12. M. Chaney, M. Buck, Mol. Microbiol. 33, 1200–1209 (1999).13. Y. Zuo, T. A. Steitz, Mol. Cell 58, 534–540 (2015).14. D. Bose et al., Mol. Cell 32, 337–346 (2008).15. K. B. Decker, D. M. Hinton,Annu. Rev. Microbiol.67, 113–139 (2013).16. G. Bar-Nahum, E. Nudler, Cell 106, 443–451 (2001).17. J. Mukhopadhyay et al., Cell 106, 453–463 (2001).18. B. Bae et al., Proc. Natl. Acad. Sci. U.S.A. 110, 19772–19777 (2013).19. A. Chakraborty et al., Science 337, 591–595 (2012).20. A. Feklistov, S. A. Darst, Cell 147, 1257–1269 (2011).21. K. S. Murakami, S. Masuda, E. A. Campbell, O. Muzzin,

S. A. Darst, Science 296, 1285–1290 (2002).22. Z. A. Chen et al., EMBO J. 29, 717–726 (2010).23. C. Engel, S. Sainsbury, A. C. Cheung, D. Kostrewa, P. Cramer,

Nature 502, 650–655 (2013).24. C. Fernández-Tornero et al., Nature 502, 644–649 (2013).25. D. Kostrewa et al., Nature 462, 323–330 (2009).26. S. Sainsbury, J. Niesser, P. Cramer, Nature 493, 437–440 (2013).27. X. Liu, D. A. Bushnell, D. Wang, G. Calero, R. D. Kornberg,

Science 327, 206–209 (2010).28. R. S. Basu et al., J. Biol. Chem. 289, 24549–24559 (2014).29. A. Sharma et al., Nucleic Acids Res. 42, 5177–5190 (2014).30. L. J. Friedman, J. Gelles, Cell 148, 679–689 (2012).

ACKNOWLEDGMENTS

We thank M. Michael and D. Bose for their earlier contributions to thisproject; A. Forster, J. Liu, and beamline scientists at the DiamondLight Source for their help with data collection; and members of X.Z.’sand M.B.’s labs for helpful discussion. We thank D. Wigley, R. Dixon,R. Weinzierl, C. Fernández-Tornero, and R. Wigneshweraraj forcritically reading the manuscript. Y.Y. was funded by the ChineseNational Science Foundation and the China Scholarship Council. Themajority of this work was funded by the UK Biotechnology andBiological Sciences Research Council to X.Z. and M.B. K.S.M. issupported by NIH grant GM087350. Y.-P.W. is funded by 973 NationalKey Basic Research Programme (2015CB755700) in China. J.T.W.and R.L.G. were supported by NIH grant R37 GM37048 (to R.L.G.).The atomic coordinate has been deposited in the Protein Data Bankwith accession code 5BYH.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/349/6250/882/suppl/DC1Materials and MethodsFigs. S1 to S5Tables S1References (31–43)

20 March 2015; accepted 15 July 201510.1126/science.aab1478

SCIENCE sciencemag.org 21 AUGUST 2015 • VOL 349 ISSUE 6250 885

Fig. 4. Regulatory elements of s54 and comparison with s70, TFIIB, and Pol I. (A) s54, (B) s70, and(C) TFIIB in relation to template strand (gold), as defined in the complex with a full transcription bubble(4YLN). (D to F) Comparison of s54 with s70, Pol I (4C2M), and TFIIB (4BBS) in the RNAP DNA-RNAchannel; structures are superposed on the bridge helix. s54 represented as spheres; s70, TFIIB, and Pol Iexpander are shown as ribbons; and RNAP as cylinders.

RESEARCH | REPORTSon June 22, 2020

http://science.sciencemag.org/

Dow

nloaded from

reveal new and conserved regulatory strategies54σStructures of the RNA polymerase-

Katsuhiko S. Murakami, Martin Buck and Xiaodong ZhangYun Yang, Vidya C. Darbari, Nan Zhang, Duo Lu, Robert Glyde, Yi-Ping Wang, Jared T. Winkelman, Richard L. Gourse,

DOI: 10.1126/science.aab1478 (6250), 882-885.349Science 

, this issue p. 882Sciencetemplate DNA from entering the RNAP active site and the downstream DNA channel.

54 blocks theσ54 complex, the σ54, as well as promoter DNA. In the initial inhibited state of the RNAP-σRNAP bound to determined the x-ray crystal structure ofet al.54 being critical for transcription of many stress response genes. Yang

σfor many cell and developmental processes. In bacteria, sigma factors play a vital role in transcription regulation, with Transcription of all genes is carried out by RNA polymerase (RNAP). The enzyme is thus a pivotal regulation point

Keeping gene transcription in check

ARTICLE TOOLS http://science.sciencemag.org/content/349/6250/882

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2015/08/19/349.6250.882.DC1

REFERENCES

http://science.sciencemag.org/content/349/6250/882#BIBLThis article cites 43 articles, 12 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 © 2015, American Association for the Advancement of Science

on June 22, 2020

http://science.sciencemag.org/

Dow

nloaded from