RNA polymerase III subunit architecture andimplications for open promoter complex formationChih-Chien Wua,b, Franz Herzogc, Stefan Jennebachd, Yu-Chun Lina, Chih-Yu Paia, Ruedi Aebersoldc,e, Patrick Cramerd,and Hung-Ta Chena,1

aInstitute of Molecular Biology, Academia Sinica, Taipei, Taiwan 115, Republic of China; bDepartment of Life Sciences and Institute of Genome Sciences,National Yang-Ming University, Taipei, Taiwan 112, Republic of China; cDepartment of Biology, Institute of Molecular Systems Biology, EidgenössicheTechnische Hochschule Zurich, CH-8093 Zurich, Switzerland; dGene Center and Department of Biochemistry, Center for Integrated Protein Science Munich,Ludwig-Maximilians-Universität München, 81377 Munich, Germany; and eFaculty of Science, University of Zurich, CH-8006 Zurich, Switzerland

Edited by E. Peter Geiduschek, University of California at San Diego, La Jolla, CA, and approved October 3, 2012 (received for review July 11, 2012)

Transcription initiation by eukaryotic RNA polymerase (Pol) III relieson the TFIIE-related subcomplex C82/34/31. Here we combine cross-linking and hydroxyl radical probing to position the C82/34/31subcomplex around the Pol III active center cleft. The extendedwinged helix (WH) domains 1 and 4 of C82 localize to the polymerasedomains clamp head and clamp core, respectively, and the two WHdomains of C34 span the polymerase cleft from the coiled-coil regionof the clamp to the protrusion. The WH domains of C82 and C34apparently cooperate with other mobile regions flanking the cleftduring promoter DNA binding, opening, and loading. Together withpublished data, our results complete the subunit architecture of PolIII and indicate that all TFIIE-related components of eukaryotic andarchaeal transcription systems adopt an evolutionarily conservedlocation in theupper part of the cleft that supports their functions inopen promoter complex formation and stabilization.

RNA polymerase III (Pol III) is the largest eukaryotic RNApolymerase, composed of 17 subunits with a total molecular

weight of∼0.7MDa (1). Pol III synthesizes certain small untranslatedRNAs (e.g., tRNAs, 5S rRNA, U6 snRNA, and 7SL RNA) involvedinRNAprocessing and translation and in protein translocation (2, 3).Human Pol III mutations have been implicated in a neurodegener-ative disorder, hypomyelinating leukodystrophy (4–6).The three eukaryotic RNA polymerases share a similar 12-sub-

unit core as represented by the X-ray structure of yeast Pol II (1). Inaddition to the core, Pol III contains five specific subunits formingtwo subcomplexes, C37/53 and C82/34/31. The C37/53 subcomplexparticipates in promoter opening, transcription termination, andpolymerase reinitiation (7–9). The C34 subunit of the C82/34/31subcomplex plays a role inopen complex formation and in recruitingPol III to the preinitiation complex (PIC) through interaction withTFIIB-related factor 1 (Brf1) (10–13). The human RPC62/39/32subcomplex, homologous to the yeast C82/34/31 complex, is disso-ciable and required for promoter-specific initiation (11).The twoPol III-specific subcomplexes contain structural domains

homologous to domains in the Pol II transcription factors TFIIFand TFIIE, including the TFIIF-related dimerizationmodule in theC37/53 subcomplex and several TFIIE-related winged helix (WH)domains in subunits C82 and C34 (14–20). TFIIE is composed oftwo subunits, Tfa1 and Tfa2, in yeast, or TFIIEα and TFIIEβ inhumans. Whereas Tfa1 bears an extended WH (eWH) domain inits N-terminal region, Tfa2 has two adjacent WH domains (21, 22).Two adjacent Tfa2-relatedWHdomains are also present in theC34subunit and its human homologRPC39. Pol I contains theA49/34.5subcomplex that features a TFIIF-like dimerization module anda tandemWH domain that contains two Tfa2-like WH folds in theC-terminal region of the A49 subunit (18). The crystal structure ofthe human C82 homolog RPC62 contains four Tfa1-like eWHdomains (eWH1–4) that are structurally organized around a C-terminal coiled-coil stalk (20). ATfa1-related eWHdomain is alsopresent in the archaeal transcription factor E (TFE) (22).Recent cryo-EM studies on the overall structural organization

of the yeast Pol III suggested locations of the Pol III-specific

subcomplexes and their contacts with the 12-subunit polymerasecore. The C37/53 dimerization module was positioned into theelectron density adjacent to the lobe domain of the C128 subuniton one side of the polymerase cleft, similar to the localization ofthe TFIIF dimerization module and the A49/34.5 dimerizationmodule on Pol II and Pol I, respectively (23–25). This structuralarrangement was supported by site-specific photo-cross-linkingand hydroxyl radical probing analyses that provided direct posi-tional mapping for protein interactions (8).The C82/34/31 subcomplex was proposed to occupy a large

electron density region between the clamp and the stalk of thepolymerase core, on the side of the cleft opposite the lobe (20, 24,25). By fitting the crystal structure of hRPC62 into the yeast PolIII cryo-EM envelope, Lefèvre et al. proposed a model in whicheWH2, eWH3, and the coiled-coil region are positioned on theclamp and eWH1 and eWH4 are exposed to solvent for single-stranded DNA binding (20). An alternative orientation wasproposed by Fernández-Tornero et al. that placed eWH4 closerto the stalk of the polymerase core (26). Thus, it currentlyremains unclear how C82 and the C82/34/31 subcomplex are po-sitioned on Pol III. To resolve this issue, protein–protein inter-actions between the C82/34/31 subcomplex and the Pol III coremust be mapped.Here we site-specifically inserted photo-cross-linking amino

acids into C160 and C82 and identified protein regions involved inC82-polymerase core interactions by site-directed hydroxyl radicalprobing. Our data reveal that C82 is anchored on the Pol III clampvia its domains eWH1 and eWH4. Mass spectrometric (MS)analysis of peptides obtained by tryptic digestion of lysine–lysinecross-linked Pol III is consistent with these results, and additionallylocates the coiled-coil stalk of C82 at the polymerase clamp–stalkjunction near the ABC27 (Rpb5) and ABC23 (Rpb6) subunits ofthe Pol III core. Cross-linking-MS analysis also positions the twoadjacent WH domains of C34 above the active site cleft. Theresulting detailed interaction map completes the subunit archi-tecture of Pol III and provides insights into the function of theC82/34/31 subcomplex during transcription initiation.

ResultsC82 and C34 Reside on the Polymerase Clamp. To elucidate proteininteractions for the Pol III active center cleft, we replaced surfaceresidues of the C160 clamp core and clamp head domains with thenonnatural amino acid photo-cross-linker p-benzoyl-L-phenylala-nine (BPA). The sites were chosen at presumed interfaces of the

Author contributions: C.-C.W., S.J., R.A., P.C., and H.-T.C. designed research; C.-C.W., F.H.,S.J., Y.-C.L., and C.-Y.P. performed research; C.-C.W., F.H., and S.J. analyzed data; and C.-C.W.,P.C., and H.-T.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: htchen@imb.sinica.edu.tw.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1211665109/-/DCSupplemental.

19232–19237 | PNAS | November 20, 2012 | vol. 109 | no. 47 www.pnas.org/cgi/doi/10.1073/pnas.1211665109

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020

Pol III core with the C82/34/31 subcomplex based on previous PolIII models (24, 25, 27). We conducted photo-cross-linking by usingyeast whole-cell extracts containing C160-BPA derivatives in theimmobilized template assay to probe potential protein targetswithin the Pol III PIC. C160-cross-linked polypeptides were foundto have estimated molecular weights of 80 and 30 kDa, corre-sponding approximately to the sizes of C82 and C34 or C31. Weepitope-tagged the possible candidates and confirmed the cross-linked polypeptides by identifying the epitope-containing cross-linked fusion in Western blot analysis, as described (8). Fig. 1Ashows Western blot analysis revealing C160–C82 cross-linksidentified by antibodies against epitope-tagged C160 and C82(anti-Myc for C160 and anti-Flag for C82).This analysis reveals C82 cross-linking to C160 residues spanning

from the upstream clamp core to the downstream clamp head (Fig.

1C; summarized in Table S1). Additionally, we found that fourconsecutive residues (Asp304 to Ile307) in the coiled-coil motif ofthe clamp core yielded a cross-link to C82 and another cross-link toC34 (Fig. 1 B and C). Cross-linking to C34 is also located fromAla206 to Pro209 in the clamp head, andHis207 and Asn208 showssimultaneous cross-linking to C82 (Fig. 1C). Taken together, ourcross-linking results indicate that the C82/34/31 subcomplex ismainly anchored on the clamp domain of C160 via its C82 subunit,and that C34 contributes to contacts along the rim of the clamp.

C82 Is Involved in a Protein Interaction Network. Our C160 cross-linking results indicate that C82 and C34 reside on the clampdomain but did not suffice to position the domains of thesesubunits on the Pol III core. We therefore conducted an extensiveBPA cross-linking analysis for C82 that is summarized in Fig. S1Aand Table S2. The data reveal C82–C31 and C82–C34 intra-subcomplex cross-links. Representative C82–C31 cross-links areshown in Fig. S1B. C31-cross-linked residues mostly lie within theC82 eWH1 and eWH2 domains, and one resides in its coiled-coildomain (Fig. S1A). By contrast, C34-cross-linked residues arescattered in eWH3 and eWH4, except for one in eWH2 (Fig. S1A and C). This C82–C34 binding interface is consistent with thepreviously proposed C82–C34 interaction obtained by proteinpull-down experiments (20). C82 shows distinct interaction sur-faces for C31 and C34, which suggests a mainly linear connectivityfor the three subunits, consistent with a previous mass spectro-metric analysis (28).Concerning the interaction of C82 with the Pol III core, res-

idues in eWH1, eWH4, the first insertion of eWH3, and the C-terminal coiled-coil stalk yield C82–C160 cross-links (Fig. 2A and

Fig. 1. C160 BPA photo-cross-linking indicates C82 and C34 reside on theC160 clamp domain. (A) Western blot of C82–C160 cross-links. Amino acidpositions in C160 clamp replaced by BPA are indicated above the lanes. C160and cross-linking bands were visualized by probing with anti-Myc antibody(C160-Myc, Left), and the cross-linking bands are confirmed to be C160–C82fusion by probing Flag-tagged C82 (Right). C160 WCE, C160 whole-cell ex-tract; UV + or −, with or without UV irradiation; WT, wild-type C160 with noBPA replacement. (B) C34-C160 cross-links. C160 and C34–C160 cross-linkingbands were visualized by anti-Myc antibody (Left) and anti-HA antibody toreveal N-terminally HA-tagged C34 (Right). As indicated, these C160-BPAderivatives also generate simultaneous C82-C160 cross-links. (C) Positions inC82– and C34–C160 cross-links in polymerase clamp. The yeast Pol III core-C37/53 surface model is colored white. Magenta sphere: magnesium ion inthe polymerase active site. As indicated, C160 residues yielding C82 and C34cross-links are shown in red and brown, respectively. The residues cross-linking to both C82 and C34 are colored mauve.

Fig. 2. C82 BPA photo-cross-linking. (A) Western blot of C82–C160 cross-linking. BPA-substituted residues in C82 are indicated above. C82 and cross-linking fusion bands are revealed by anti-V5 antibody (Left). The Flag-epitopetagged C160 in the cross-linking bands is revealed by anti-Flag antibody(Right). Asterisks indicate unidentified cross-linked polypeptides. (B) Aminoacid positions cross-linked to C160 in the yeast C82 model. The yeast C82homology model (displayed as the ribbon model of Cα trace) is generated bythe Modeler program (42) using the human RPC62 structure (PDB 2XUB) (20)as the template. The dashed line indicates the missing residues (aa547–567 inC82) in the eWH4. C82 structural domains are colored in red (eWH1), green(eWH2), blue (eWH3), orange (eWH4), and salmon (coiled-coil). The colorscheme is used for all following figures. Residues cross-linking to C160 andC25 are highlighted with cyan and olive spheres, respectively.

Wu et al. PNAS | November 20, 2012 | vol. 109 | no. 47 | 19233

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020

Fig. S1A). By mapping the positions cross-linked to C160 on theyeast C82 homology model based on the human RPC62 struc-ture, we reveal a protein interaction surface for C160 bindingthat comprised eWH1, eWH4, and the coiled-coil stalk of C82(Fig. 2B). In addition to C160 and intrasubcomplex cross-links,Leu620-BPA near the tip of C82 coiled-coil showed simulta-neous cross-links to C25 of the polymerase stalk and C53 of theC37/53 subcomplex (Fig. S1 A and D). This result is consistentwith our previous data indicating that the C53 N terminusextends toward the polymerase stalk to allow for cross-linking toC82 and C25 (8).

C82 eWH1 and eWH4 Reside on the Polymerase Clamp. To moreprecisely localize C82 domains on the C160 clamp domain, weapplied directed hydroxyl radical probing analysis with C82-FeBABE conjugates. We purified a series of recombinant C82single-cysteine variants to tether the ·OH-generating FeBABEreagent at defined Cys positions. These C82-FeBABE conjugateswere combined with purified recombinant C34 and C31 to re-constitute the C82/34/31 subcomplex (purified proteins are shown

in Fig. S2A). The reconstituted subcomplex was subsequently ti-trated into a C82mutant yeast whole-cell extract (Fig. S2B) for PolIII PIC formation with the immobilized template assay. AllFeBABE-tethered C82 variants are shown to restore transcriptionactivity (Fig. S2C). Fig. 3A shows hydroxyl radical cleavage of C160from individual C82 variants with FeBABE tethered at Cys posi-tions in the eWH1 and eWH4 domains. By mapping the cleavagesites to the Pol III core structural model, we found that eWH1 andeWH4 are respectively localized to the clamp head and clamp coreregions (Fig. 3B; summarized in Table S3).

Cross-Linking-MS Analysis of Pol III.AlthoughBPAcross-linking andhydroxyl radical probing localized C82 eWH1/4 on the C160clamp, additional structural restraints are required to further de-fine the interface between C82 and the Pol III core for structuralmodeling. To this end, we used lysine–lysine cross-linking followedby MS identification of the cross-linked peptides as recently de-scribed for Pol I (23). A purified endogenous Pol III sample wassubjected to cross-linking with disuccinimidyl suberate (DSS),a reagent that reacts with primary amines present in lysine side

Fig. 3. Localization of C82 on the C160 clamp. (A) Directed hydroxyl radical cleavage of C160. Western blots of C160 cleavage from tethered FeBABE in C82eWH1 and eWH4 are respectively shown in the left and right panels. C-terminally HA-tagged C160 fragments are revealed by anti-HA antibody. FeBABEpositions in C82 are indicated above the lanes. Cleavage fragments in the lower-molecular-weight range are displayed with signals amplified (Bottom Right).Red asterisks mark identified cleavage fragments. Approximate locations of cleavage sites in C160 are indicated. (B) C82-Pol III core model. C82 (ribbon model)is manually docked as a rigid body on the Pol III core model (white surface model) based on directed hydroxyl radical cleavage and cross-linking-MS analyses.Hydroxyl radical cleavage regions in C160 clamp are highlighted in black covering 11 residues centered at the deduced cut site. (C) Cross-linked lysine pairsbetween C82 coiled-coil and subunits of the Pol III core. Green lines connect cross-linked lysine pairs from C82 (Lys594) to ABC27 (Lys171) and ABC23 (Lys72).The BPA substitution at C82 Leu620, highlighted as an olive sphere, cross-links to C25.

19234 | www.pnas.org/cgi/doi/10.1073/pnas.1211665109 Wu et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020

chain and protein N terminus. DSS cross-linking of Pol III wasmonitored by tryptic digestion and MS analysis. A complete list ofcross-linked lysines and their corresponding peptide pairs is pro-vided in Table S4.The cross-linking identified linkage pairs that validate a previous

model of the Pol III core that was based on the Pol II structure (27,29). In addition, distance restraints position the Pol III-specificsubcomplex C37/53 on the Pol III core, supporting the previouslyproposed model for the C37/53 dimerization module on the lobedomain of C128 (Fig. S3 A and B) (8, 24, 25, 29). Cross-links alsoreveal the location of the C-terminal region of C37 in the activecenter cleft (Fig. S3B) and are generally consistent with contactsbetween the N-terminal region of C53 and the coiled-coil domainof C82 and the C17/25 stalk subcomplex (Fig. S1D), supportingprevious BPA cross-linking and FeBABE analyses (8).

Model for C82 Binding to the Pol III Core. A schematic summary forC82 lysine–lysine cross-links is depicted in Fig. S4. We observedintramolecular cross-links to support the yeast C82 homologymodel based on the human RPC62 structure (29). In combinationwith the C82-FeBABE hydroxyl radical analysis (see above), weused intermolecular cross-links between C82 coiled-coil and thePol III core subunits ABC27 (Rpb6) and ABC23 (Rpb5) tomanually dock the C82 model on the Pol III core (Fig. 3 B andC).In the resulting model, the C82 domains eWH1 and eWH4 arelocated on the clamp head and clamp core in C160, respectively,whereas the C82 coiled-coil extends toward subunits ABC27 andABC23. C82 coiled-coil is also positioned adjacent to the Pol IIIstalk, consistent with BPA cross-linking between the tip of C82coiled-coil and C25 of the Pol III stalk (Fig. 3C). Further, themodel satisfies all BPA photo-cross-linking results for C160 andC82 (Fig. S4 B and C).To validate our model of C82 on the Pol III core, we generated

yeast strains containing mutations in the predicted C160 clamp–C82 interface and tested their in vivo and in vitro phenotypes. Weobtained a deletion mutant C160 Δ (35–41) that shows a slow-growth phenotype at elevated temperature but no significantchange in the cellular C160 protein level (Fig. 4A, Input). In animmunoprecipitation assay by immobilizing C82, the C34 and C31subunits of the subcomplex are coimmunoprecipitated, whereasthe mutant C160 and another Pol III subunit C53 are partiallydissociated (Fig. 4A, Flag C82 IP). By analyzing PIC formation onimmobilized SUP4 tDNA, C160 Δ (35–41) mutant showed ∼50%reduction inPol III recruitment (Fig. 4A, PIC andB). In agreementwith the defect in Pol III recruitment, we observed a concomitantdecrease in transcription initiation for the C160 mutant based onthe single-round transcription analysis (Fig. 4 A Lower and B).Further mutagenesis study on C82 allowed us to isolate two

internal deletion mutations in C82 eWH4 that confer tempera-ture-sensitive growth defect. Coimmunoprecipitation analysis byimmobilizing the Pol III core subunit C128 indicates that thesemutations lead to dissociation of C82 and C34 from the Pol IIIcore (Fig. 4C). This association defect severely compromises PolIII recruitment and causes complete loss of in vitro transcriptionactivity (Fig. 4 C and D), in agreement with a role for the humanC82/34/31 homolog in transcription initiation (11). Taken to-gether, our mutational and functional analyses support theproposed model for C82 binding to the Pol III core.

C34 Bridges the Active Center Cleft. In addition to C82 positioning,we used lysine–lysine cross-linked peptide pairs to model the posi-tions for C34 WH domains. Homology models for C34 WH1 andWH2 were generated based on the WH structures in mouse andhuman RPC39, respectively (Protein Data Bank accession numbers2DK8 and 2DK5). Based on a total of eight intersubunit cross-links,we docked the C34WH2model above the Pol III active center cleftto contact C128 protrusion, C160 clamp coiled-coil, and the eWH4domain of the previouslymodeledC82 (Fig. 5).With the positioning

for WH2, the adjacent WH1 is localized on the rim of the C160clamp head based on five intrasubunit cross-links with WH2 anda single intersubunit cross-link with C82 eWH4. The positions forWH1 and WH2 are consistent with the localization of C34 on thepolymerase clamp based on the BPA cross-linking results for C160(Figs. 5 and 1C). Further, several cross-link pairs are observed be-tween the C34 WH domains and the C37 C-terminal region, sup-porting the previously described location for the C37 C-terminalregion near the Pol III active site (8) (Fig. S3C).Intersubunit cross-linked peptide pairs within the C82/34/31

subcomplex further suggest the positions of the C34 C-terminalregion and C31 on C82. The C34 C-terminal region contacts C82eWH3, and C31 is positioned on the surface formed by C82 eWH1,eWH2, and the coiled-coil stalk (Fig. S5 A and B). The C31 posi-tion is also supported by cross-links with the C160 clamp head, theC17/25 stalk, and ABC27 (Fig. S5A). Furthermore, the distinctlocation of C34 and C31 and no observable cross-links betweenC34 and C31 are in perfect agreement with the linear connectivity

Fig. 4. Association of the C82/34/31 subcomplex with the Pol III core. (A)Western blot analyses of coimmunoprecipitation and Pol III PIC formationwith yeast whole-cell extract containing a C160 mutation. Coimmunopreci-pitation was conducted with anti-Flag agarose to immobilize Flag-epitope-tagged C82 (Flag C82 IP; Upper Left). Pol III PIC formation assay was con-ducted with the immobilized SUP4 tDNA (Upper Middle). The level of a TFIIICsubunit Tfc4 in the PIC is also shown. As indicated, whole-cell extracts con-taining wild-type (WT) or internal deletion Δ (35–41) C160 were used. Single-round in vitro transcription of the isolated Pol III PIC was assayed (Lower). (B)Immunoblot and in vitro transcription signals were quantified from threeindependent experiments with WT signals set to 1. Error bars indicate SD. (C)Western blot analyses of coimmunoprecipitation and Pol III PIC formationwith yeast whole-cell extracts containing C82 mutations. Similar to theanalyses in A, whole-cell extracts containingWT or the indicated mutations inC82 eWH4 were used in coimmunoprecipitation, PIC formation, and in vitrotranscription analyses. Different from the analysis in A, immune pull-downwas conducted by immobilizing Flag-epitope-tagged C128, and both single-round (SR) and multiple-round (MR) transcription were assayed (Lower). Anasterisk indicates the read-through transcripts. (D) The results for PIC for-mation and SR transcription from C are quantified and plotted with WTsignals set to 1. Errors bars indicate SD from three independent experiments.

Wu et al. PNAS | November 20, 2012 | vol. 109 | no. 47 | 19235

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020

for the C82/34/31 subcomplex indicated by our C82 BPA-cross-linking results and a previous MS study (28). Taken together, themultiple experimental approaches have enabled us to arrive at thecomplete domain architecture for the 17-subunit Pol III enzyme.

DiscussionHere we present the complete subunit domain architecture of PolIII, the largest eukaryotic RNA polymerase, derived from an in-tegrative approach that combines nonnatural amino acid photo-cross-linking, directed hydroxyl-radical probing, MS analysis oflysine–lysine chemically cross-linked peptides, and molecularmodeling based on X-ray crystallographic structures. This workidentified subunit–subunit interfaces within Pol III, and in par-ticular revealed the location of subcomplex C82/34/31 around thePol III clamp and cleft. The C82 domains eWH1 and eWH4 arelocalized on the clamp head and clamp core, respectively, whereasthe C82 C-terminal coiled-coil domain binds between the poly-merase clamp and stalk and reaches near the subunits ABC27 andABC23. This model differs from previously proposed relativeorientations of the C82 human homolog structure on Pol III basedon electron microscopic data alone (20, 24, 25). The C34 domainWH2 resides between the C160 coiled-coil motif on one side ofthe polymerase cleft and the C128 protrusion domain on the otherside, whereas the C34 domain WH1 localizes to the clamp head.C31 is positioned adjacent to the clamp head and contacts C82eWH1, eWH2, and its coiled-coil domain, ABC27, and the C17/25stalk. The subunit domain model of Pol III derived here is gen-erally consistent with the molecular envelope of the Pol III EMstructure (24, 25) (Fig. S6 A and B).To derive implications for the function of C82 andC34 in Pol III

initiation, we extended the Pol III model to a model of the Pol IIIopen promoter complex (Fig. 6). The open promoter complexmodel additionally contains the TATA box-binding protein TBPand the TFIIB-related factor Brf1 based on previous analysis ofthe related Pol II open complex (37). This model is consistent withpreviously reported protein–DNA cross-linking results, in whichC34 was cross-linked to the farthest upstream position at−21 withrespect to the transcription start +1, and C82 was cross-linkedboth upstream at positions −8/−7 and downstream at +11 (Fig.6B) (13, 38). The C34 location correlates well with its double-stranded DNA binding ability and its functional roles in DNAmelting (12, 20). With their positions flanking the DNA bubble ofthe Pol III open complex model (Fig. 6B), the C34 WH domainspotentially contribute their DNA binding residues to function in

initial strand separation and/or subsequent bubble stabilization.In accordance with this proposed mechanism, two amino acidresidues in theWH2 domain of C34 that were previously reportedto be involved in DNA binding and open complex formation (12,20) may interact with the upstream edge of the transcriptionbubble (Fig. 6B). Our data and published results thus converge ona model that C34 is positioned over the active center cleft tostabilize theDNAbubble and aid in open complex formation. Thismodel relies on C34 mobility, because promoter DNA must firstbe loaded into the cleft during the transition from the closed tothe open complex, and this requires transient displacement ofC34 from the location observed here. Weak evidence to supportC34 mobility comes from the slightly different locations for C34domains in Pol III and Pol III-DNA models derived by EM (20,24, 25) but merits further exploration.C82 and C34 may cooperate with the TFIIB-related factor Brf1

to stimulate open complex formation. In our Pol III model, C82eWH4 and C34WH domains are in the vicinity of the polymeraseclamp coiled-coil (Fig. 6A and Fig. S6C). In Pol II, the clampcoiled-coil interacts with a region in TFIIB, the B-linker, whichwas also implicated in DNA opening (37, 39). During Pol IIIinitiation, Brf1 likely contacts C82 and C34, because C34 wasshown to bind Brf1 during Pol III recruitment (10, 40). Situated atthe upstream edge of the DNA bubble (Fig. 6B), C82, C34, andBrf1 may provide essential protein–DNA interactions for initialspontaneous strand separation, similar to the function of σ2/3 inthe bacterial system (30, 37, 39, 41). In particular, the C34 WHdomains and C82 eWH4 could stabilize the emerging non-template DNA strand, whereas the template DNA strand slipsinto the cleft and binds to its floor (31, 32, 37). On the downstreamside of the DNA bubble, C82 eWH1 could contact downstreamDNA (Fig. 6A). Further, the C37 C-terminal domain and anothertranscription factor, Bdp1, may also bind the DNA bubble (8, 33),because they localize near the Pol III active center (Fig. S3C).Our results suggest an evolutionary conservation of the mech-

anisms used by TFIIE-related components in the three eukaryoticand the archaeal transcription initiation machineries. In Pol II,

Fig. 5. Model of C34 WH domains in Pol III based on lysine–lysine cross-links.Model of C34 and C82 on the Pol III core is shown on the left. C34WH1 andWH2ribbonmodels are colored purple. In the Pol III cleft viewon the right, green linesconnect intrasubunit and intersubunit cross-linked lysines (purple spheres withlabels indicating subunit names and residue numbers). Brown patches on C160clamp indicate the BPA-substituted residues in C160 that cross-linked to C34.

Fig. 6. Pol III architecture and open promoter complex. (A) Model for thePol III open promoter complex. The model of Pol III-Brf1-TBP-DNA openpromoter complex was built based on the Pol II-TFIIB-TBP open complex (37)and the Brf1-TBP-DNA structure (43). Template and nontemplate DNAstrands are in blue and cyan, respectively. The Pol III core-C37/53 model isdisplayed as the white surface model. Ribbon models for Brf1 and TBP arecolored yellow-green and brown, respectively. The C82 surface model iscolored according to the domain color scheme as above, and both C34 WHsurface models are colored purple. (B) Protein–DNA organization in the PolIII active center. The Pol III open promoter complex in A is displayed withdifferent orientation to view the Pol III active center. The Pol III core issemitransparent and C34 is changed to the ribbon model. Lys135 and Lys138in C34 WH2 are highlighted as black spheres. These two amino acids arerequired for DNA binding and open complex formation (12, 20).

19236 | www.pnas.org/cgi/doi/10.1073/pnas.1211665109 Wu et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020

TFIIE has been mapped to the coiled-coil motif and the clamphead of Rpb1 (34, 35). In Pol I, the TFIIE-related A49 tandemwinged-helix domain occupies a position above the active centercleft similar to C34 (23). The archaeal TFIIE-like factor TFE wasalso localized near the clamp coiled-coil (36). Thus, all TFIIE-related components share the location within the upper activecenter cleft and a function in promoting formation and stabiliza-tion of the transcription bubble.

Materials and MethodsDetailed descriptions are available in SI Materials and Methods for plasmidsand yeast strains, BPA photo-cross-linking, preparation of proteins, in vitro

transcription, hydroxyl radical cleavage, and cross-linking-MS analysis.Immobilized DNA templates containing either SUP4 tRNA or U6 snRNAgenes were prepared as previously described (8). The Pol III core-C37/53model was built as previously described (8).

ACKNOWLEDGMENTS. We thank Cheng-Feng Lo and Jin-Cheng Lee forassistance with cloning and development of purification strategies for C82,C34, and C31. This work was supported by National Science Council Grant100-2311-B-001-013-MY3 and a Career Development Award (to H.-T.C.) fromAcademia Sinica. P.C. was supported by Deutsche ForschungsgemeinschaftGrant SFB646, TR5, Nanosystems Initiative Munich, the BioImaging Network,a European Research Council Advanced Grant, and a Jung-Stiftung grant.F.H. is supported by a European Molecular Biology Organization long-termfellowship and by European Commission Grant FP7-PEOPLE-IEF.

1. Cramer P, et al. (2008) Structure of eukaryotic RNA polymerases. Annu Rev Biophys37:337–352.

2. Dieci G, Fiorino G, Castelnuovo M, Teichmann M, Pagano A (2007) The expandingRNA polymerase III transcriptome. Trends Genet 23(12):614–622.

3. Schramm L, Hernandez N (2002) Recruitment of RNA polymerase III to its targetpromoters. Genes Dev 16(20):2593–2620.

4. Bernard G, et al. (2011) Mutations of POLR3A encoding a catalytic subunit of RNApolymerase Pol III cause a recessive hypomyelinating leukodystrophy. Am J HumGenet 89(3):415–423.

5. Tétreault M, et al. (2011) Recessive mutations in POLR3B, encoding the second largestsubunit of Pol III, cause a rare hypomyelinating leukodystrophy. Am J Hum Genet89(5):652–655.

6. Saitsu H, et al. (2011) Mutations in POLR3A and POLR3B encoding RNA Polymerase IIIsubunits cause an autosomal-recessive hypomyelinating leukoencephalopathy. Am JHum Genet 89(5):644–651.

7. Kassavetis GA, Prakash P, Shim E (2010) The C53/C37 subcomplex of RNA polymeraseIII lies near the active site and participates in promoter opening. J Biol Chem 285(4):2695–2706.

8. Wu CC, Lin YC, Chen HT (2011) The TFIIF-like Rpc37/53 dimer lies at the center ofa protein network to connect TFIIIC, Bdp1, and the RNA polymerase III active center.Mol Cell Biol 31(13):2715–2728.

9. Landrieux E, et al. (2006) A subcomplex of RNA polymerase III subunits involved intranscription termination and reinitiation. EMBO J 25(1):118–128.

10. Werner M, Chaussivert N, Willis IM, Sentenac A (1993) Interaction between a complexof RNA polymerase III subunits and the 70-kDa component of transcription factor IIIB.J Biol Chem 268(28):20721–20724.

11. Wang Z, Roeder RG (1997) Three human RNA polymerase III-specific subunits forma subcomplex with a selective function in specific transcription initiation. Genes Dev11(10):1315–1326.

12. Brun I, Sentenac A, Werner M (1997) Dual role of the C34 subunit of RNA polymeraseIII in transcription initiation. EMBO J 16(18):5730–5741.

13. Bartholomew B, Durkovich D, Kassavetis GA, Geiduschek EP (1993) Orientation andtopography of RNA polymerase III in transcription complexes. Mol Cell Biol 13(2):942–952.

14. Carter R, Drouin G (2010) The increase in the number of subunits in eukaryotic RNApolymerase III relative to RNA polymerase II is due to the permanent recruitment ofgeneral transcription factors. Mol Biol Evol 27(5):1035–1043.

15. Knutson BA, Hahn S (2011) Yeast Rrn7 and human TAF1B are TFIIB-related RNApolymerase I general transcription factors. Science 333(6049):1637–1640.

16. Naidu S, Friedrich JK, Russell J, Zomerdijk JC (2011) TAF1B is a TFIIB-like component ofthe basal transcription machinery for RNA polymerase I. Science 333(6049):1640–1642.

17. Werner F, Grohmann D (2011) Evolution of multisubunit RNA polymerases in thethree domains of life. Nat Rev Microbiol 9(2):85–98.

18. Geiger SR, et al. (2010) RNA polymerase I contains a TFIIF-related DNA-binding sub-complex. Mol Cell 39(4):583–594.

19. Vannini A, Cramer P (2012) Conservation between the RNA polymerase I, II, and IIItranscription initiation machineries. Mol Cell 45(4):439–446.

20. Lefèvre S, et al. (2011) Structure-function analysis of hRPC62 provides insights intoRNA polymerase III transcription initiation. Nat Struct Mol Biol 18(3):352–358.

21. Luo J, Fishburn J, Hahn S, Ranish J (2012) An integrated chemical cross-linking andmass spectrometry approach to study protein complex architecture and function. MolCell Proteomics 11(2):M111.008318.

22. Meinhart A, Blobel J, Cramer P (2003) An extended winged helix domain in generaltranscription factor E/IIE alpha. J Biol Chem 278(48):48267–48274.

23. Jennebach S, Herzog F, Aebersold R, Cramer P (2012) Crosslinking-MS analysis revealsRNA polymerase I domain architecture and basis of rRNA cleavage. Nucleic Acids Res40(12):5591–5601.

24. Vannini A, et al. (2010) Molecular basis of RNA polymerase III transcription repressionby Maf1. Cell 143(1):59–70.

25. Fernández-Tornero C, et al. (2010) Conformational flexibility of RNA polymerase IIIduring transcriptional elongation. EMBO J 29(22):3762–3772.

26. Fernández-Tornero C, Böttcher B, Rashid UJ, Müller CW (2011) Analyzing RNA poly-merase III by electron cryomicroscopy. RNA Biol 8(5):760–765.

27. Jasiak AJ, Armache KJ, Martens B, Jansen RP, Cramer P (2006) Structural biology ofRNA polymerase III: Subcomplex C17/25 X-ray structure and 11 subunit enzymemodel. Mol Cell 23(1):71–81.

28. Lane LA, et al. (2011) Mass spectrometry reveals stable modules in holo and apo RNApolymerases I and III. Structure 19(1):90–100.

29. Jennebach S (2011) RNA polymerase I domain architecture and basis of rRNA cleav-age. PhD dissertation (Ludwig-Maximilians-Universität München, Munich).

30. Feklistov A, Darst SA (2011) Structural basis for promoter-10 element recognition bythe bacterial RNA polymerase σ subunit. Cell 147(6):1257–1269.

31. Naji S, Bertero MG, Spitalny P, Cramer P, Thomm M (2008) Structure-function analysisof the RNA polymerase cleft loops elucidates initial transcription, DNA unwinding andRNA displacement. Nucleic Acids Res 36(2):676–687.

32. Cheung AC, Sainsbury S, Cramer P (2011) Structural basis of initial RNA polymerase IItranscription. EMBO J 30(23):4755–4763.

33. Kassavetis GA, Letts GA, Geiduschek EP (2001) The RNA polymerase III transcriptioninitiation factor TFIIIB participates in two steps of promoter opening. EMBO J 20(11):2823–2834.

34. Chen HT, Warfield L, Hahn S (2007) The positions of TFIIF and TFIIE in the RNApolymerase II transcription preinitiation complex. Nat Struct Mol Biol 14(8):696–703.

35. Grünberg S, Warfield L, Hahn S (2012) Architecture of the RNA polymerase II pre-initiation complex and mechanism of ATP-dependent promoter opening. Nat StructMol Biol 19(8):788–796.

36. Grohmann D, et al. (2011) The initiation factor TFE and the elongation factor Spt4/5compete for the RNAP clamp during transcription initiation and elongation. Mol Cell43(2):263–274.

37. Kostrewa D, et al. (2009) RNA polymerase II-TFIIB structure and mechanism of tran-scription initiation. Nature 462(7271):323–330.

38. Kassavetis GA, Han S, Naji S, Geiduschek EP (2003) The role of transcription initiationfactor IIIB subunits in promoter opening probed by photochemical cross-linking. J BiolChem 278(20):17912–17917.

39. Liu X, Bushnell DA, Wang D, Calero G, Kornberg RD (2010) Structure of an RNApolymerase II-TFIIB complex and the transcription initiation mechanism. Science 327(5962):206–209.

40. Khoo B, Brophy B, Jackson SP (1994) Conserved functional domains of the RNApolymerase III general transcription factor BRF. Genes Dev 8(23):2879–2890.

41. Artsimovitch I, et al. (2005) Allosteric modulation of the RNA polymerase catalyticreaction is an essential component of transcription control by rifamycins. Cell 122(3):351–363.

42. Eswar N, Eramian D, Webb B, Shen MY, Sali A (2008) Protein structure modeling withMODELLER. Methods Mol Biol 426:145–159.

43. Juo ZS, Kassavetis GA, Wang J, Geiduschek EP, Sigler PB (2003) Crystal structure ofa transcription factor IIIB core interface ternary complex. Nature 422(6931):534–539.

Wu et al. PNAS | November 20, 2012 | vol. 109 | no. 47 | 19237

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

7, 2

020