TRANSCRIPTION

Structural basis oftranscription activationYu Feng, Yu Zhang, Richard H. Ebright*

Class II transcription activators function by binding to a DNA site overlapping acore promoter and stimulating isomerization of an initial RNA polymerase (RNAP)–promoter closed complex into a catalytically competent RNAP-promoter opencomplex. Here, we report a 4.4 angstrom crystal structure of an intact bacterial class IItranscription activation complex. The structure comprises Thermus thermophilustranscription activator protein TTHB099 (TAP) [homolog of Escherichia colicatabolite activator protein (CAP)], T. thermophilus RNAP sA holoenzyme, a class IITAP-dependent promoter, and a ribotetranucleotide primer. The structure reveals theinteractions between RNAP holoenzyme and DNA responsible for transcriptioninitiation and reveals the interactions between TAP and RNAP holoenzymeresponsible for transcription activation. The structure indicates that TAP stimulatesisomerization through simple, adhesive, stabilizing protein-protein interactions withRNAP holoenzyme.

Simple bacterial transcription activators—those that stimulate transcription from asingle DNA site without other factors—aredivided into two classes (1–3). Class I tran-scription activators, typified by Escherichia

coli catabolite activator protein (CAP) at the lacpromoter, stimulate transcription by binding toa specific DNA site upstream of a core promoterand facilitating binding of RNA polymerase(RNAP) holoenzyme to form an RNAP-promoterclosed complex (RPc) (1–3). Class II transcriptionactivators, typified by E. coli CAP at the gal pro-moter, stimulate transcription by binding to aspecific DNA site overlapping a core promoterand facilitating conversion of RPc into a cata-lytically competent RNAP-promoter open com-plex (RPo) containing ~13 base pairs (bp) ofunwound DNA (“transcription bubble”) (1–3).A 20 Å–resolution electron microscopy struc-ture of a class I transcription activation complexhas been reported (4), but no structure of a classII transcription activation complex previouslyhas been reported. Here, we determine the4.4 Å–resolution crystal structure of a class IItranscription activation complex comprisingThermus thermophilus transcription activatorprotein TTHB099 (TAP) (a thermophilic se-quence, structural, and functional homolog ofE. coli CAP) (5); T. thermophilus RNAP sA holo-enzyme; a class II TAP-dependent promoter;and the ribotetranucleotide primer UpCpGpA(TAP-Rpo) (table S1, Fig. 1, and figs. S1 and S2).To obtain a structure of TAP-RPo, we used a

nucleic-acid scaffold corresponding to positions–57 to +15 of a class II TAP-dependent promoter(positions numbered relative to transcription start

site) (Fig. 1A and fig. S1). The scaffold contained aconsensus DNA site for TAP centered betweenpositions –41 and –42 (same position as DNA sitefor E. coli CAP in gal promoter) (1, 2), a near-consensus extended –10 element (3), a consensus–10 element (3), a consensus discriminator element(3), a consensus core recognition element (3), a13-bp transcription bubble (maintained in the un-wound state by having noncomplementary se-quences on nontemplate and template strands),and UpCpGpA.In the structure of TAP-RPo, TAP interacts with

DNA, RNAP holoenzyme interacts with DNA, andTAP and RNAP holoenzyme make protein-proteininteractions (Fig. 1, B and C). The structure of TAP-DNA in TAP-RPo is superimposable on the struc-ture of CAP-DNA (6), corroborating that TAP is ahomolog of CAP (Fig. 1D). The structure of RPoin TAP-RPo is essentially superimposable on struc-tures of RPo (7–10) [neglecting RNAP a subunit C-terminal domain (aCTD), which was not resolvedin previous structures], indicating that interactionsbetween the class II activator and RPo do not sub-stantially alter the conformation of RPo (Fig. 1E).RNAP contains two copies of aCTD, each of

which is connected to the rest of RNAP througha flexible linker (1–3). In the structure of TAP-RPo, one aCTD (probably aCTDI) (fig. S3) inter-acts with TAP, and the other aCTD (probablyaCTDII) (fig. S3) makes no interactions (Fig. 1, Band C). In the crystal, the second aCTD is con-strained by lattice contacts (i.e., contacts with TAPin an adjacent molecule of TAP-RPo in the lattice)(fig. S4). In solution, this aCTD would be free toadopt other positions.The structure defines the interactions between

RNAP holoenzyme and DNA that mediate pro-moter recognition and promoter unwinding intranscription initiation (Figs. 1 and 2) and the inter-actions between TAP and RNAP holoenzyme thatmediate transcription activation (Figs. 1, 3, and 4).

TAP and s conserved region sR4 “corecognize”the promoter –35 region, contacting the sameDNA segment from different faces of the DNAhelix (Figs. 1, B and C, and 2A). The general modeof interaction of sR4 with –35-region DNA inTAP-RPo—binding of the second a helix of thesR4 helix-turn-helix motif in the DNA majorgroove—is the same as in RPo (Fig. 2A and fig.S5) (8–10), but, due to DNA distortion by TAP,–35-region DNA is rotated ~20° away from sR4(fig. S5). This rotation decreases the number ofsR4 residues that contact DNA bases from 3 to2 and decreases the number of contacted DNAbases from 4 to 2, providing a structural explana-tion for the observation that, although –35-regionDNA sequences are recognized in class II activator-dependent transcription, the recognition specificityis less and the number of recognized bases issmaller than in activator-independent transcription(11). Two sR4 residues are positioned to makecontacts with DNA bases that potentially enablesequence readout (Fig. 2A). Substitution of theseresidues reduces RPo formation, verifying theirimportance (Fig. 2A).s conserved region sR3 interacts with the

promoter extended –10 region (Figs. 1, B andC, and 2B) (8–10). Three sR3 residues are posi-tioned to make contacts with DNA bases (Fig.2B and fig. S6). Substitution of these residuesreduces RPo formation, verifying their impor-tance (Fig. 2B).s conserved region sR2 interacts with the

promoter –10 element at the “upstream forkjunction” where DNA unwinding occurs to formthe transcription bubble (Figs. 1E and 2C).sR2 interacts with the first position of the –10element (–12) as double-stranded DNA (dsDNA)and the second through sixth positions of the–10 element (–11 through –7) as nontemplate-strand single-stranded DNA (ssDNA) (Figs. 1Eand 2C). sR2 Trp433 (numbered as in E. coli s70)is positioned to stack on the nontemplate-strandbase of base pair –12, forming a “wedge” thatforces the nontemplate-stand –11 base to unstackand flip outside the DNA helix (9), where it iscaptured by binding within a pocket formedby residues of sR2 (Fig. 2C and fig. S6) (7–10).sR2 Arg436 is positioned to stack on the template-strand base of base pair –12, forming an anal-ogous “wedge” that forces the template-strand –11base to unstack and flip outside the DNA helix,where it is captured within a channel formed byresidues of RNAP, sR2, and sR3.2 that leadsinto the RNAP active-center cleft (Fig. 2C andfigs. S6 and S7). Substitution of Trp433 or Arg436

results in defects in RPo formation, verifying theirimportance (Fig. 2C). A second pair of residues,Gln437 and Thr440, are positioned to make directcontacts with the nontemplate- and template-strand bases of the –12 base pair, providing astructural explanation for the observation thatsubstitution of these residues alters specificityat –12 (Fig. 2C and fig. S6) (12).s conserved region sR1.2 interacts with

nontemplate-strand ssDNA of the discriminatorelement, as described previously (Fig. 1E) (7–10).RNAP core interacts with the nontemplate-strand

1330 10 JUNE 2016 • VOL 352 ISSUE 6291 sciencemag.org SCIENCE

Waksman Institute and Department of Chemistry and ChemicalBiology, Rutgers University, Piscataway, NJ 08854, USA.*Corresponding author. Email: ebright@waksman.rutgers.edu

RESEARCH | REPORTSon A

ugust 26, 2020

http://science.sciencemag.org/

Dow

nloaded from

ssDNA of the core recognition element, template-strand ssDNA of the transcription bubble, anddownstream dsDNA, as described previously (Fig.1E) (7–10).Genetic and biochemical experiments indicate

that class II transcription activation by E. coli CAPinvolves three sets of protein-protein interactions:(i) activating region 1 (AR1) interacts with aCTD,(ii) activating region 2 (AR2) interacts with aspecies-specific insertion in RNAP aI subunitN-terminal domain (aNTDI) (162-165 determinant),and (iii) activating region 3 (AR3) interacts withsR4 (1, 2).In TAP-RPo, a surface of TAP corresponding

to AR2 approaches aNTDI and contacts the RNAPb flap (Fig. 3A). Three residues of TAP AR2 arepositioned to make direct contacts with threeresidues of RNAP b subunit (Fig. 3B). TAP Glu77

and RNAP b Arg735 are positioned to form a saltbridge in the AR2-RNAP interface (Fig. 3B). Charge-reversal substitution of either residue decreasesTAP-dependent transcription, and charge-reversalsubstitution of both residues, which recreatesa salt bridge, restores TAP-dependent transcrip-tion, confirming the importance of the inferredinteraction (Fig. 3B). Homology modeling of

CAP-RPo based on TAP-RPo indicates that CAPAR2 is positioned to contact the aNTDI 162-165 determinant (a species-specific insertion pres-ent in E. coli RNAP but not in T. thermophilusRNAP) (fig. S8, A and B), consistent with pre-vious work (13). Homology modeling indicatesthat CAP AR2 also is positioned to contact theRNAP b flap (fig. S8, A and B). Substitution ofthe inferred interacting residues decreasesCAP-dependent transcription, indicating thatthe inferred interactions occur and are impor-tant (fig. S8B).In TAP-RPo, a surface of TAP corresponding

to AR3 contacts sR4 a helices 4 and 5 and theRNAP b flap-tip a helix (Fig. 3A). Eight predom-inantly negatively charged residues of TAP AR3are positioned to interact with six predominant-ly positively charged residues of sR4 and threepredominantly positively charged residues ofthe b flap-tip a helix (Fig. 3C). TAP Glu15 is po-sitioned to form a salt bridge with a sR4 Argresidue at the center of the interface (Fig. 3C).Charge-reversal substitution of either residuedecreases TAP-dependent transcription, andcharge-reversal substitution of both residues,recreating a salt bridge, restores TAP-dependent

transcription, indicating that the interactionsoccur and are important (Fig. 3C). Homologymodeling of CAP-RPo based on TAP-RPo pre-dicts equivalent interactions between seven pre-dominantly negatively charged residues of CAPAR3 and five predominantly positively chargedresidues of sR4 and one residue of the b flap-tip(fig. S8, A and C), consistent with previous work(14, 15).In TAP-RPo, the surface of TAP correspond-

ing to AR1 makes no interactions, and, instead,a different surface of TAP, here designated “ac-tivating region 4” (AR4), interacts with aCTD(Figs. 1, B and C, and 4A). The interface betweenTAP AR4 and aCTD is large (300 Å2) (Fig. 4B).Nine residues of TAP AR4 are positioned to makedirect contacts with eight residues of aCTD (Fig.4B). Substitution of residues implicated in TAPAR4-aCTD interaction results in defects in TAP-dependent transcription (Fig. 4B). TAP-aCTD in-teractions differ from CAP-aCTD interactionsnot only in the identities of the activating re-gions (AR4 in TAP; AR1 in CAP) but also in thefact that TAP interacts with aCTD not bound toDNA, whereas CAP interacts with aCTD boundto DNA immediately upstream of CAP (Figs. 1,

SCIENCE sciencemag.org 10 JUNE 2016 • VOL 352 ISSUE 6291 1331

Fig. 1. Structure of TAP-RPo. (A) Nucleic-acid scaffold. Pink, nontemplate strand; red, template strand; magenta, UpCpGpA; violet, extended –10 element; blue,–10 element; light blue, discriminator element. (B andC) TAP-RPo [ribbons in (B); surfaces in (C); b' nonconserved region omitted for clarity].Cyan,TAP; yellow, s;white, green, gray, and dark gray, RNAP aNTD, aCTD, b, and b', respectively.Other colors as in (A). Dashed lines, aNTD-aCTD linkers. (D) Comparison of TAP-DNAin TAP-RPo [colors as in (B) and (C)] to CAP-DNA (gray) (6). (E) Comparison of transcription bubble and downstreamdsDNA in TAP-RPo [colors as in (B) and (C)]to corresponding DNA segments in RPo (cyan) (7).

RESEARCH | REPORTSon A

ugust 26, 2020

http://science.sciencemag.org/

Dow

nloaded from

B and C, and 4A) (1, 2). Hydroxyl-radical DNAfootprinting confirms that aCTD functionsdifferently in T. thermophilus than in E. coli.Thus, T. thermophilus aCTD does not footprintDNA at a class II TAP-dependent promoter ora ribosomal RNA (rRNA) promoter (figs. S9 toS11), in contrast to E. coli aCTD, which foot-prints DNA immediately upstream of CAP at aclass II CAP-dependent promoter and adenine-thymine–rich upstream-element DNA immedi-ately upstream of the –35 element at a rRNApromoter (figs. S9 to S11). Consistent with thestructure of TAP-RPo, fluorescence-polarization

assays show that TAP is able to bind to aCTDin the absence of DNA and that the bindingrequires AR4 (Fig. 4C, left). Further consistentwith the structure, fluorescence-polarizationassays show that TAP is able to bind to RNAPholoenzyme in the absence of DNA and thatthe binding requires AR4 interactions and doesnot require AR2 and AR3 interactions (Fig. 4C,right).The finding that TAP is able to bind to RNAP

holoenzyme in the absence of DNA raises thepossibility that TAP, in contrast to CAP, can accessnot only a “recruitment” pathway, in which the

activator interacts first with DNA and then withRNAP holoenzyme, but also a “prerecruitment”pathway, in which the activator interacts first withRNAP holoenzyme and then with DNA (fig. S12)(16). Based on the equilibrium dissociation con-stant (KD) for TAP-RNAP holoenzyme interaction(6 mM) (Fig. 4C, right) and the concentration ofnontranscribing RNAP in bacteria in vivo (5 mM)(17), it appears likely that a prerecruitment path-way contributes to TAP-dependent transcriptionin T. thermophilus in vivo.Measurements of effects of substitution of TAP-

activating regions on the kinetics of transcription

1332 10 JUNE 2016 • VOL 352 ISSUE 6291 sciencemag.org SCIENCE

Fig. 2. Protein-DNA interactions that mediate promoter recognition.Green, s residues that contact DNA bases (numbered as in E. coli s70);brown, sR2 residues that contact nontemplate-strand base –11. Othercolors as in Fig. 1, B and C. Graphs, effects on RPo formation of Alasubstitutions of E. coli s70 (mean ± SEM; ≥3 determinations). (A) Inter-actions between sR4 and –35 region. (B) Interactions between sR3 andextended –10 element. (C) Interactions between sR2 and nontemplate-and template-strand nucleotides of first (–12NT, –12T) and second (–11NT,–11T) positions of –10 element.

Fig. 3. Protein-protein interactions that mediate transcription activation:AR2 and AR3. Green, TAP AR2; blue, TAP AR3; orange, RNAP b-flap residuesthat contact AR2; magenta and light magenta, sR4 and RNAP b-flap-tipresidues that contact AR3 (numbered as in TAP and T. thermophilus RNAPholoenzyme and, in parentheses, as in CAP and E. coli RNAP holoenzyme).Other colors as in Fig. 1, B and C. Graphs, effects on TAP-dependenttranscription of single and double charge-reversal substitutions (mean ±SEM; ≥3 determinations). (A) Interactions between AR2, AR3, and RNAPholoenzyme (left, ribbons; right, surfaces). (B) AR2 interactions. (C) AR3interactions.

RESEARCH | REPORTSon A

ugust 26, 2020

http://science.sciencemag.org/

Dow

nloaded from

initiation indicate that TAP AR2 and AR3 promoteisomerization of RPc to RPo (kf), and TAP AR4promotes formation of RPc (KB) (Fig. 4D). Thispattern is reminiscent of the pattern for E. coliCAP, for which AR2 and AR3 promote isomer-ization of RPc to RPo (11, 13, 15), and AR1, throughinteraction with aCTD, promotes formation ofRPc (11, 13).A long-standing question has been how a

class II activator promotes isomerization ofRPc to RPo, which entails loading of DNAinto the RNAP active-center cleft, unwindingof DNA to form the transcription bubble, andclosure of the RNAP clamp (1–3, 13, 18–20). Thestructure of TAP-RPo reveals that TAP does notinteract with, and does not alter the conforma-tion or interactions of, the RNAP active-centercleft, the transcription bubble, or the RNAPclamp. The structure further reveals that theinteractions that promote isomerization—AR2and AR3 interactions—are simple, adhesive, stabi-lizing protein-protein interactions between ex-posed surfaces of TAP and exposed surfaces ofRNAP holoenzyme (Fig. 3 and fig. S8). We inferthat interactions between a class II activatorand RNAP holoenzyme that promote forma-tion of RPc (AR4 interactions for TAP; AR1 in-teractions for CAP) and interactions between

class II activator and RNAP holoenzyme thatpromote isomerization (AR2 and AR3 interac-tions) do not differ in character but, instead,differ only in timing (13, 18–20). The formerfirst occur in the transition state for formationof RPc and stabilize both RPc and RPo, whereasthe latter first occur in the transition state forisomerization of RPc to RPo and stabilize RPo(Fig. 4E).

REFERENCES AND NOTES

1. S. Busby, R. H. Ebright, J. Mol. Biol. 293, 199–213(1999).

2. C. L. Lawson et al., Curr. Opin. Struct. Biol. 14, 10–20(2004).

3. K. B. Decker, D. M. Hinton, Annu. Rev. Microbiol. 67, 113–139(2013).

4. B. P. Hudson et al., Proc. Natl. Acad. Sci. U.S.A. 106,19830–19835 (2009).

5. Y. Agari, S. Kuramitsu, A. Shinkai, Proteins 80, 1490–1494(2012).

6. S. C. Schultz, G. C. Shields, T. A. Steitz, Science 253,1001–1007 (1991).

7. Y. Zhang et al., Science 338, 1076–1080 (2012).8. Y. Zuo, T. A. Steitz, Mol. Cell 58, 534–540 (2015).9. B. Bae, A. Feklistov, A. Lass-Napiorkowska, R. Landick,

S. A. Darst, eLife 4, e08504 (2015).10. B. Bae et al., eLife 4, e08505 (2015).11. V. A. Rhodius, D. M. West, C. L. Webster, S. J. Busby,

N. J. Savery, Nucleic Acids Res. 25, 326–332 (1997).

12. C. L. Chan, M. A. Lonetto, C. A. Gross, Structure 4, 1235–1238(1996).

13. W. Niu, Y. Kim, G. Tau, T. Heyduk, R. H. Ebright, Cell 87,1123–1134 (1996).

14. M. A. Lonetto et al., J. Mol. Biol. 284, 1353–1365(1998).

15. V. A. Rhodius, S. J. Busby, J. Mol. Biol. 299, 311–324(2000).

16. M. A. Zafar, I. M. Shah, R. E. Wolf Jr., J. Mol. Biol. 401, 13–32(2010).

17. M. Patrick, P. P. Dennis, M. Ehrenberg, H. Bremer, Biochimie119, 80–91 (2015).

18. M. Ptashne, A. Gann, Nature 386, 569–577(1997).

19. S. Roy, S. Garges, S. Adhya, J. Biol. Chem. 273, 14059–14062(1998).

20. S. L. Dove, F. W. Huang, A. Hochschild, Proc. Natl. Acad. Sci. U.S.A.97, 13215–13220 (2000).

ACKNOWLEDGMENTS

We thank the National Synchrotron Light Source for beamline accessand E. Arnold for discussion. This work was funded by NIH grantGM041376 to R.H.E. The Protein Data Bank accession codeis 5I2D.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/352/6291/1330/suppl/DC1Materials and MethodsFigs. S1 to S12Table S1References (21–55)

8 February 2016; accepted 9 May 201610.1126/science.aaf4417

SCIENCE sciencemag.org 10 JUNE 2016 • VOL 352 ISSUE 6291 1333

Fig. 4. Protein-protein interactions that mediate transcription activa-tion: AR4 and kinetics. (A and B) TAP AR4 interaction with aCTD (left,ribbons; right, surfaces; lower left, close-up). Violet,TAP AR4; yellow, aCTDresidues that contact AR4. Other colors as in Fig. 1, B and C. Graph,effects on TAP-dependent transcription of charge-reversal substitutions ofAR4 and aCTD or truncation of aCTD (mean ± SEM; ≥3 determinations).(C) TAP-aCTD (left) and TAP-RNAP (right) interactions in the absence ofDNA. (D) Data (left) and parameters (right) for effects of substitutionsof AR2, AR3, and AR4 on kinetics of transcription initiation. (E) Summaryof class II activator-dependent transcription.

RESEARCH | REPORTSon A

ugust 26, 2020

http://science.sciencemag.org/

Dow

nloaded from

Structural basis of transcription activationYu Feng, Yu Zhang and Richard H. Ebright

DOI: 10.1126/science.aaf4417 (6291), 1330-1333.352Science 

, this issue p. 1330Scienceformation is critical for activation.RNAP active center or the RNAP clamp. Instead, it seems that the timing of the interaction during transcription complex complex through simple stabilizing protein-protein interactions with RNAP. The critical contacts did not go through theactivation complex. The transcription activator protein (TAP) converts the closed RNAP-promoter complex into an open

determined the crystal structure of a bacterial transcriptionet al.factors influence the activity of the RNAP. Feng Regulating transcription by RNA polymerase (RNAP) is central to controlling gene expression. Transcription

Transcription activation all about timing

ARTICLE TOOLS http://science.sciencemag.org/content/352/6291/1330

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2016/06/08/352.6291.1330.DC1

REFERENCES

http://science.sciencemag.org/content/352/6291/1330#BIBLThis article cites 54 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 © 2016, American Association for the Advancement of Science

on August 26, 2020

http://science.sciencem

ag.org/D

ownloaded from