Proc. Natl. Acad. Sci. USAVol. 92, pp. 7026-7030, July 1995Biochemistry

Structure and function of a human transcription factor TFIIIBsubunit that is evolutionarily conserved and contains bothTFIIB- and high-mobility-group protein 2-related domainsZHENGXIN WANG AND ROBERT G. ROEDER*

The Rockefeller University, New York, NY 10021

Contributed by Robert G. Roeder, April 24, 1995

ABSTRACT Transcription factor TFIIIB plays a centralrole in transcription initiation by RNA polymerase III ongenes encoding tRNA, 5S rRNA, and other small structuralRNAs. We report the purification of a human TFIIIB-derivedcomplex containing only the TATA-binding polypeptide (TBP)and a 90-kDa subunit (TFIIIB90) and the isolation of a cDNAclone encoding the 90-kDa subunit. The N-terminal half ofTFIIIB90 exhibits sequence similarity to the yeast TFIIIB70(BRF) and the class II transcription factor TFIIB and inter-acts weakly with TBP. The C-terminal half of TFIIIB90contains a high-mobility-group protein 2 (HMG2)-relateddomain and interacts strongly with TBP. RecombinantTFIIIB90 plus recombinant human TBP substitute for humanTFIIIB in a complementation assay for transcription of 5S,tRNA, and VAl RNA genes, and both the TFIIB-relateddomain and the HMG2-related domain are required for thisactivity. TFIIIB90 is also required for transcription of human7SK and U6 RNA genes by RNA polymerase Ill, but apparentlywithin a complex distinct from the TBP/TFTIIB90 complex.

Transcription of small structural RNA genes by RNA poly-merase (pol) III involves a number of accessory factors (forreview, see refs. 1 and 2). These include the "common" factorsTFIIIC and TFIIIB, which suffice for transcription of tRNAand virus-associated (VA) RNA genes, and, in some cases,additional gene-specific factors (e.g., TFIIIA for 5S genes anda proximal sequence element-binding factor for mammalianU6 and 7SK genes). Transcription of tRNA and VA RNAgenes begins with recognition of internal promoter elements(A and B boxes) by TFIIIC, followed by sequential binding ofTFIIIB and pol III. The process is similar for 5S RNA genes,except that TFIIIC recruitment is dependent upon priorbinding of TFIIIA to a distinct set of internal promoterelements (3). Studies in yeast have indicated a primary role forTFIIIB as an initiation factor, since it can remain stably bound(in a position-specific but sequence-independent manner) andfacilitate pol III recruitment and function following dissocia-tion of TFIIIC or TFIIIC plus TFIIIA from the correspondingpreinitiation complexes (4). TFIIIB is also of special interest,since, the TATA-binding polypeptide (TBP) commonly re-quired by all nuclear RNA polymerases (for review, see ref 5)is an essential component. TFIIIB structure and function arebest understood in yeast, where additional subunits of 70 kDa(BRF, TFIIIB70) and 90 kDa (TFIIIB90) account, with TBP,for all the properties of yeast TFIIIB (6). Studies have dem-onstrated sequential binding ofyeast TFIIIB70, TBP, and yeastTFIIIB90 to a TFIIIC promoter complex (6), as well asaccompanying conformational changes in TFIIIC and TFIIIBcomponents (6, 7). In agreement with these findings and therole of TFIIIB in pol III recruitment, yeast TFIIIB70 wasshown to interact independently with a yeast TFIIIC subunit,

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

with TBP, and with a pol III subunit (6-11). Consistent withparallel roles of TFIIIB and TFIIB in the recruitment of polIII and pol II, respectively, the N-terminal domain of yeastTFIIIB70 is related in sequence to TFIIB (11-14). The indi-cation that TFIIB also serves as a target for various gene-specific activators (15) suggests the possibility of a similar rolefor TFIIIB on class III genes.Apart from TBP, the components of mammalian TFIIIB are

unclear. This reflects in part the presence of distinct TBPcomplexes, as well as pol III and a subcomplex (TFIIICl-related) of TFIIIC, in the classical TFIIIB fraction (unpub-lished data; see also ref. 2). Thus, in this TFIIIB fraction acombination of conventional chromatographic and affinity puri-fication procedures using antibodies that recognize TBP haveidentified TBP-associated components of (i) 150, 82, and 54 kDa(16), (ii) 190, 97, 87, and 60 kDa (17), and (iii) 172 kDa and anundefined size [TBP-associated factor L (TAF-L)] (18). Giventhat the two mammalian TFIIIC subunits characterized to dateshow no sequence relationship to yeast TFIIIC subunits (forreview, see ref 19) and that mammalian class III genes withexternal promoter elements, unlike their yeast counterparts,require additional gene-specific factors (see ref. 2), a greatercomplexity in mammalian TFIIIB structure might be expected.To examine these questions we have further purified human

TFIIIB and here report the cloning and characterization of a90-kDa TFIIIB subunit (TFIIIB90)t having both similaritiesand differences with the yeast TFIIIB70 subunit.

MATERIALS AND METHODS

Purification of TFIIIB. One hundred milliliters (1200 mg ofprotein) of HeLa cell S100 fraction (20) was loaded onto aphosphocellulose (Whatman P11) column (3.5 cm x 10 cm) inbuffer C [20mM Hepes, pH 7.9/0.2mM EDTA/10% (vol/vol)glycerol/0.5 mM phenylmethylsulfonyl fluoride/3 mM dithio-threitol] containing 100 mM KCl (BC100), and, after washing,step-eluted with BC350 and BC600 (21). The BC350 fraction (200mg of protein) was diluted with BCO to 200mM KCl, loaded ontoa Mono Q HR10/10 FPLC column (Pharmacia), and eluted witha 20-column-volume linear gradient from BC200 to BC600. Forsome experiments, Mono Q factions containing TFIIIB activitywere pooled and incubated with antigen-purified anti-TBP anti-body-protein A-Sepharose 4B for 3-6 hr. The resin was washedwith BC300/0.1% Nonidet P-40 and bound material was elutedwith 100mM glycine at pH 2.5. Eluents were neutralized with 1.5M Tris HCl at pH 8.8 or by dialysis against BC100 for 3 hr to

Abbreviations: BCn, buffer C with n mM KCl; GST, glutathioneS-transferase; HMG2, high-mobility-group protein 2; TBP, TATA-binding polypeptide; TAF, TBP-associated factor; pol III (II), RNApolymerase III (II).*To whom reprint requests should be addressed at: Laboratory ofBiochemistry and Molecular Biology, The Rockefeller University,Box 268, 1230 York Avenue, New York, NY 10021.IThe sequence reported in this paper has been deposited in theGenBank data base (accession no. U28838).

7026

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

Proc. Natl. Acad. Sci. USA 92 (1995) 7027

renature proteins. The native TFIIIB (HPLC SP-5PW fraction)was purified as described (17).

Recombinant TFIIIB90 and Deletion Mutants. The codingregion of the full-length TFIIIB90 clone (see text) was FLAGepitope-tagged and subcloned into baculovirus expressionvector pVL1392 (17) for expression in Sf9 insect cells asdirected (PharMingen). Expressed protein was purified byimmunoprecipitation with anti-FLAG monoclonal antibodyM2 resin (17). Deletion clones pETB90(1-300), pETB90(1-461), and pETB90(281-675) were constructed by subcloningcorresponding PCR fragments into the bacterial expressionvector pETl5d (22). The corresponding hexahistidine-taggedproteins TFIIIB90-(1-300) and TFIIIB90-(281-675) were ex-pressed in bacteria and purified through Ni2+-NTA-agarose(Qiagen, Chatsworth, CA) (22). For the protein-protein in-teraction studies, truncated TFIIIB90 cDNA fragments weresubcloned into pGEX-2LT(+) vector [derived from pGEX-2T(Pharmacia); A. Hoffmann and R.G.R., unpublished data] andthe corresponding glutathione S-transferase (GST) fusionproteins were expressed in bacteria (22).

RESULTSPurification of TFIIIB. TFIIIB was purified from HeLa

S100, since this fraction contains nearly as much TFIIIBactivity as does the nuclear extract, but very little TFIID andB-TFIID. The conventional 0.3 M KCI phosphocellulose frac-tion containing TFIIIB (21) was subjected to Mono Q FPLC,which resulted in the separation of TFIIIB (eluted at 0.28 MKCl) and pol III (eluted at 0.35 M KCl). HPLC of the MonoQ TFIIIB fraction on SP-5PW resulted in a complete separa-tion from pol III, as well as contaminating TFIIIC. An anti-gen-purified anti-TBP polyclonal antibody was used to immu-noprecipitate a TBP-containing complex from the Mono QTFIIIB fraction. Somewhat surprisingly, the resulting complexcontained only TBP and two 90-kDa polypeptides (Fig. 1A).Both of the 90-kDa polypeptides bound 35S-labeled TBP in aprotein blot analysis (Fig. 1B). The polypeptides eluted fromthe anti-TBP column were renatured by dialysis against BC100for 3 hr and assayed in a complementation assay with highlypurified pol III (24), partially purified TFIIIC, and a VAlRNA template (Fig. 1C). In conjunction with recombinanthuman TBP (lane 4), but not in the absence of exogenous TBP

A B C

97 *j_

i |BgI 1

66*l a

4 TBP45 ,'="f.

+ + + + + Pol lII+ + + + TFIIIC

+ + + ren IIIB+ rTBP

+ TFIIIB

(lane 3), the renatured polypeptides showed a high activity,comparable to that partially purified TFIIIB (lane 5). Theseresults indicate that the polypeptides eluted from the anti-TBPantibody column had TFIIIB activity.

Cloning of a cDNA Encoding TFIIIB90. TFIIIB purifiedfrom -3 liters of HeLa S100 fraction (36,000 mg of protein)was subjected to SDS/7% PAGE and the resolved polypep-tides were transferred to a poly(vinylidene difluoride) mem-brane. The region containing TFIIIB90 (two bands) wasexcised and digested with endoproteinase Lys-C. Eluted pep-tides were separated by reverse-phase HPLC and subjected tomicrosequence analysis. Peptides EEVEGEISSYQDAIEIEand GDGELDLSGIDDLEID were used to design oligonu-cleotides 5'-GAIGAIGTIGAIGGIGARATITCITCITAYG-ARGAYGCIATIGARATIGA-3' and 5'-GGIGAYGGI-GARITIGAYYTIIIIGGIATIGAYGAYITIGARATIGA-3'.Screening of a Namalwa (Burkitt lymphoma) cell AZAP1cDNA library yielded a positive clone with a 2-kb insert thatcontained both probe sequences. Screening of additionalhuman cell (K562 erythroleukemia, 293 embryonic kidney,and HeLa) cDNA libraries with the 5' end of this insert yieldeda 3662-bp DNA fragment with a 675-aa open reading frame.This open reading frame (Fig. 2A) encodes a highly charged(30.8% acidic plus basic residues) polypeptide with a calcu-lated mass of 74,380 Da and a predicted isoelectric point of 6.75.The derived protein sequence contains all four peptide sequencesobtained by microsequence analysis (underlined in Fig. 2A).

Structural Features of TFIIIB90. Comparison of theTFIIIB90 sequence with available data bases (PDB, GenBank,and EMBL; April 1995) revealed a striking sequence relation-ship (30% identity and 53% similarity overall) to the yeastTFIIIB70 (BRF) (11-14), with much greater conservation inthe N-terminal half (41% identity, 61% similarity) than in theC-terminal half (12.5% identity, 31% similarity) (GCG pro-

AMTGRVCRGCGG 1ID I ET,DAA DI VCSEACG VEDOiVEVQFVES SGGG SAVGQFVSI.DGG(Z.KT7'LCGC Gr G ESRAQTLQDGRR IiIIlGNQLQLjNQHCLDT'tI NF- KMAVSRFHLTRGRFKMAHVI AACCLYLVCRTECSTPIIMLLVLVSDL-LQVNVYJVTLGCKTF L LLAPF -X.CTICA IY3 I 3 P-CLLi j PR FAS l1 L EE FG3EKNHEVSMTA.LiR.LLQRMl. PDWMHTGRG PS GI !CGGALLVAARMHDFRRT-7KEv7i SVKE' -L L RKI-'TEFEDC2PTSQLTIDEFMK iDLEEECDPP S Y'L'IAGQF-KRTv,KQ FQJVL SKKLEEVEGEISSYODAIEIEL,.N'PS'RE)KR(l-G 0'(,F'FKR'.L RS;r'RVQLVWR

OQP QP C PWV(Lx, (UKP7'CG HiR C G S RG ;)r:-~.QJr~.,ir( l,lltt- QL SSRK PR JtGRQ1T ALGSLTLDPLPb"ACAS.T SD.'TXFC I S SQS SDPKDASDGELDTSGiDLIDRV IDL1j-E AR1- LWMRENAEYL

REOREKEARIAKEKELGIYKEHK PKKJSCKRREPIQASI'AREATIE KMLEQTKK L S S.XTNY r-C SSAGGGS PHREDAQ P E

rlSARKLSPPI~~~~~~~~~TIL v77RTV7'TQKH-S1 SA -7AKL S RFR, X- P,,c, -1..RV'i(;' G Kj1- ;nRT T VSTQPANKK;VA2TCEA~LL7JSPTLC,AE;ARt-)AVLV'C'Gi,VYYllADEFSADEEEPDEEDG~i(>CzJ'.QMr:l 2'-t ,:'V IC-ED)DG'

B

A -.a VA1

1 100 200 300 400 500

yllIB70 -s-

t 41% (61 %)

4080

12016020024()2503201 6 U

6 8064 04. 4,

6020

t12.5% (31%)

: 1 2 3 4 S

FIG. 1. (A) SDS/7% PAGE of anti-TBP antibody-purified TFIIIB(5 ng). The gel was stained with silver. Positions and sizes (kDa) ofmarker proteins run in parallel are at left. (B) Protein blot of TFIIIB.Purified TFIIIB (5 ng) was subjected to SDS/7% PAGE, transferredto a membrane, and probed with 35S-labeled TBP. Bound TBP wasrevealed by autoradiography (23). (C) In vitro transcription assays. Thebasic complementation assay mixture (24) contained 2.4 ,ug of partiallypurified TFIIIC (0.6 M KCl P11 fraction), 0.1 ,ul of highly purified polIII, and a VA1 template (lane 1). Recombinant TBP (rTBP, 5 ng), 2utkl of the renatured immunopurified TFIIIB, or 2 t,l of the chromato-graphically purified TFIIIB (HPLC SP-SPW fraction; see ref. 17) wasadded as indicated.

_ 21%(45%)hiuB -F _ _

20% (43.9%)

HMG2Box 1 Box 1 Acidic

FIG. 2. Structure of human TFIIIB90. (A) Predicted amino acidsequence. The four peptide sequences obtained by microsequenceanalysis are underlined. (B) Schematic of structural motifs and se-quence relationships of human TFIIIB90 (hIIIB90) to yeast TFIIIB70(yIIIB70; refs. 11-14), human TFIIB (hIIB; ref. 25), and chickenhigh-mobility group protein 2 (HMG-2; ref. 26). In hIIIB90 thepositions of the indicated motifs are as follows: zinc finger (ZF), aa1-30; direct repeats (arrows), aa 96-171 and 186-270; HMG boxes(Box 1 and Box 2), aa 436-501 and 502-585; acidic tail, aa 585-675.Percent amino acid sequence identities (and similarities) are shown.

17CI I I1-11 I U.. I

Biochemistry: Wang and Roeder

niiiuuu VRz-_I..... --1111112on II

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

7028 Biochemistry: Wang and Roeder

gram, Genetics Computer Group, Madison, WI). Similar tothe situation for yeast TFIIIB70 and TFIIB, and as summa-rized in Fig. 2B, the N-terminal half (aa 1-295) of humanTFIIIB90 is related in sequence (21% identity, 45% similarity)to the human RNA polymerase II transcription factor TFIIB(for review, see ref. 25). In contrast to the situation for yeastTFIIIB70, and as indicated in Fig. 2B, the divergent and highlycharged C-terminal half of human TFIIIB90 contains anHMG2-related region (aa 436-651; 20% identity, 43.9% sim-ilarity with chicken HMG2); it also shows sequence relation-ships (aa 264-589; 21% identity, 40% similarity) with theVolvox carteri histone H1-I. The HMG2-related region (aa436-651) also shows similarity with HMG box-containinghuman UBF (27). While the somewhat low sequence similaritybetween the C-terminal region of TFIIIB90 and HMG2 pre-cludes firm conclusions about its significance, the acidic C-terminal tail (aa 585-675) of TFIIIB90 is typical of HMG2proteins and the HMG2-related region has a predicted helix-turn-helix-turn-helix structure (data not shown) similar tothat reported for HMG2 (28, 29). HMG1, HMG2, and relatedproteins contain two HMG boxes and recognize DNA with noobvious, or with low, sequence specificity (for review, see ref.30). Like yeast TFIIIB70 and all eukaryotic TFIIBs (forreview, see ref. 25), human TFIIIB90 contains a putative zincfinger at the N terminus (aa 1-30) followed by two directrepeats (aa 96-171 and 187-270). A potential leucine zipperstructure is also present in human TFIIIB90 (aa 149-170).

Expression of Functional TFIIIB90 from a Cloned cDNA.Recombinant FLAG-tagged (17) TFIIIB90 expressed in Sf9cells with a baculovirus vector migrated as a 92-kDa protein(s)(Fig. 3A) and also bound TBP in a Western blot assay (Fig. 3B).Recombinant TFIIIB90 plus recombinant human TBP (Fig.3C, lane 6), but not either alone (lanes 2 and 3), fully sub-stituted for chromatographically purified natural TFIIIB (lane7) in a complementation assay using highly purified pol III,partially purified TFIIIC, and the VAl gene. No transcriptionwas observed in the absence of TFIIIB (Fig. 3C, lane 1),TFIIIC (lane 4), or pol III (lane 5). These results show that theisolated cDNA clone encodes the 90-kDa subunit of TFIIIBand that the TBP/TFIIIB90 complex is functionally equivalentto traditionally defined TFIIIB (see Discussion).Both Conserved N-Terminal and Nonconserved C-Ter-

minal Parts of TFIIIB90 Interact with TBP and Are Essential

for Transcription. Consistent with the copurification of TBPand TFIIIB90 during chromatography and immunoprecipita-tion, and the results of Fig. 1B, protein blot analysis revealeddirect interactions between TBP and recombinant TFIIIB90(Fig. 3B; Fig. 4B, lane 4). To elucidate the region(s) ofTFIIIB90 involved in the interaction with TBP, TBP bindingto N- and C-terminal halves of TFIIIB90 was analyzed. Proteinblot analysis with a TBP probe showed strong binding to theC-terminal half (aa 281-675) (Fig. 4B, lane 6) but no detect-able binding to the TFIIB-related N-terminal half (aa 1-300)(Fig. 4B, lane 5). However, binding assays with immobilizedFLAG-tagged or GST fusion proteins (Fig. 4C; see also figurelegend) showed a weak interaction (resistant to 100 mM KClbut sensitive to 0.1% Nonidet P-40 plus 100 or 300 mM KCl)of TBP with the N-terminal half of TFIIIB90 (Fig. 4C, lane 10vs. lane 8), as well as a strong interaction (resistant to 0.1%Nonidet P-40 plus 300 mM KCl) with the C-terminal half ofTFIIIB90 (lane 11 vs. lane 8) or with the intact protein (lane9 vs. lane 7). These observations are in agreement with reportsof yeast TBP interactions with intact yeast TFIIIB70 (10) aswell as both N- and C-terminal domains (11). That a TBPmutant which specifically inhibits pol III transcription alsoreduces an interaction with the C-terminal part of yeastTFIIIB70 attests to the biological significance of these C-terminal interactions (11).For functional analyses, the intact and truncated TFIIIB90

proteins were expressed in a rabbit reticulocyte lysate, purifiedon Ni2+-NTA-agarose (Fig. SA, lanes 1-3), and assayed in a

A -% B I,, C ,,

C7; v 0 X-CM CM_N_CM

0) o00 000o o F @q00

mmX m mXX c mm ma== ~ -=-== o

97*

66*

45* V-S W --TBP

A B C220*

116*

97* 11IB90-_ _

+ + + + + + Po III+ + + + + + TFIIIC

+ + + + rilIB90+ + + + rTBP

+ TFIIIB

66*

_VAl

45*

1 2 3 4 5 6 7

FIG. 3. Function of recombinant human TFIIIB90. (A) SDS/PAGE analysis of purified baculovirus-expressed TFIIIB90 (5 ng) as

in Fig. LA. (B) Protein blot of baculovirus-expressed TFIIIB90 (5 ng)probed with 35S-labeled TBP as in Fig. LB. (C) In vitro transcriptionwith recombinant TFIIIB90 and TBP. The basic complementationassay mixture (legend to Fig. 1) with the VAI gene contained TFIIIC,pol III, recombinant TBP (rTBP, 5 ng), recombinant TFIIIB90(rIIIB90, 5 ng), and chromatographically purified TFIIIB (2 ,ul) as

indicated.

32* " 7 8 9 1011

1 2 3 4 5 6

FIG. 4. Interactions of TBP with TFIIIB90. (A) Samples (100 ng)of recombinant TFIIIB90, TFIIIB90(1-300), and TFIIIB90(281-675)proteins were resolved by SDS/7% PAGE and stained with Coomassieblue. Lane M shows low-range protein molecular weight markers(Bio-Rad) with sizes (kDa) indicated at left. The recombinantTFIIIB90(281-675) protein contains a degradation product migratingat the same position as TFIIIB90(1-300) (lane 3). (B) Protein blot ofrecombinant TFIIIB90 (10 ng) and deletion mutants of TFIIIB90 (10ng) (lanes 4-6) probed with 35S-labeled TBP as in Fig. 1B. (C)Interactions of TBP with TFIIIB90 fusion proteins. Anti-FLAGmonoclonal antibody M2 resin (10 ,ul) with (lanes 9) or without (lanes7) bound full-length FIAG-TFIIIB90 or glutathione beads (10 ,ul)bearing GST alone (lanes 8) or GST-truncated TFIIIB90 fusionproteins (lanes 10 and 11) were incubated at 4°C for 1 hr on a rotatingwheel with 0.1 ,ug of recombinant TBP. The amounts (1 jig) ofTFIIIB90 and truncated TFIIIB90 proteins were normalized by SDS/PAGE with Coomassie blue staining. For lanes 7-11, the incubationand wash buffer was BC100 (Top), BC100/0.1% Nonidet P-40 (Mid-dle), on BC300/0.1% Nonidet P-40 (Bottom). The washed beads wereboiled in 50 ,l of SDS sample buffer and 5-,ul samples were loadedonto an SDS/10% polyacrylamide gel and analyzed by Westernimmunoblot with anti-human TBP antibody.

Proc. Natl. Acad. Sci. USA 92 (1995)

4M.110''M -.-m- TBP

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

Proc. Natl. Acad. Sci. USA 92 (1995) 7029

system reconstituted with the VAl gene, recombinant TBP,pol III, and partially purified TFIIIC (Fig. SB, lanes 4-7).Deletion of either the N-terminal, TFIIB-related half (lane 7)or the C-terminal, HMG2-related domain (lane 6) totallyabolished TFIIIB transcription activity, whereas intactTFIIIB90 was fully active at an equivalent molar ratio (lane 5vs. lane 4). These results indicate that both domains areimportant for TFIIIB90 function, and are consistent with theobservation that the C terminus of yeast TFIIIB70 is requiredfor cell viability (14).U6 and 7SK Gene Transcription Requires a Distinct Com-

plex Containing TBP and TFIIIB90. Although a role for TBPin human 7SK and U6 transcription by pol III is established,the involvement of TAFs is not clear (16, 31). To test the roleof human TFIIIB90, nuclear extract was incubated with pre-immune serum or a corresponding antiserum to humanTFIIIB90, a procedure which removed >90% of the endoge-nous TFIIIB90 (data not shown; see ref. 19). As expected,anti-TFIIIB90 antiserum, but not preimmune serum, dramat-ically reduced VA1, tRNA, and 5S RNA transcription and,somewhat surprisingly, both U6 and 7SK transcription (Fig. 6,lanes 1 and 2). However, while either native TFIIIB (lane 3)or recombinant TBP plus recombinant TFIIIB90 (lane 4)almost completely restored transcription of VA1, tRNA, and5S RNA genes (with added TFIIIA in the latter case), neitherhad any detectable effect on 7SK and U6 transcription. Thesestudies establish a role for TFIIIB90 in 7SK and U6 transcrip-tion but also suggest that the corresponding complex differs insome way from that involved for VA1 transcription.

DISCUSSIONThe central role played by TFIIIB in the recognition of marked(activator/accessory factor-bound) class III promoters and inthe subsequent recruitment of pol III (see Introduction)necessitates an investigation of its structure and function. Herewe report the cloning of a cDNA encoding a TFIIB-relatedsubunit of human TFIIIB and a further analysis of its structureand function in the transcription of distinct class III genes.Composition and Function of a Mammalian TFIIIB. Using

an assay system composed of pol III and a partially purified

A UnY-W TL

I-

_co

m m In_-_ _

_)0_0_

B

_a

t (0_rl CM

m In m_-_ _

_ _ _

1 23 4 56 7

FIG. 5. N-terminal, TFIIB-related and C-terminal, HMG2-related

domains are required for transcription. (A) Rabbit reticulocyte lysate-translated TFIIIB90 and de'letion mutants. The in vitro translated

proteins, labeled with [35S]methionine, were purified on Ni2+-NTA-agarose by virtue of their hexahistidine tags and analyzed by SDS/8%PAGE followed by autoradiography. (B) In vitro transcription with

reticulocyte lysate-translated T'FIIIB90 and deletion mutants. The

basic complementation assay mixture (legend to Fig. 1) with the VAl

gene contained TFIIIC, pol III, and TBP (lane 4) with addition of

full-length TFIIIB90 (5 ng) or deletion mutants (5 ng) as indicated

(lanes 5-7).

9

(((K~~~~~~~~~~~~~Al;

5S _ I,

tRNA _ 3

VA1 _:

7SK ~=1

So

U6 -_.

1 2 3 4 5 6

FIG. 6. Involvement of TFIIIB90 in transcription of various classIII genes. In vitro transcription from genes encoding human 5S rRNA(nt -85 to + 164 in plasmid pUC19; see ref. 32), human tRNAl et (33),adenovirus VAl RNA human 7SK RNA (2), and mouse U6 RNA (2)was assayed with preimmune serum-depleted nuclear extract (lane 1)or anti-TFIIIB90 antibody-depleted nuclear extract (lanes 2-6) withno addition (lane 2) or with recombinant (rT13P and rTFIIIB90) ornatural [TFIIIB and TFIIIA (32)] factors added as indicated.

TFIIIC (conventional 0.6 M KCl phosphocellulose fractioncontaining both TFIIIC1 and TFIIIC2; see ref. 16), we havepurified a minimal TFIIIB complex containing TBP and a90-kDa polypeptide. That TBP and the 90-kDa species repre-sent active TFIIIB components is verified by the ability ofrecombinant TBP and recombinant TFIIIB90 to fully substi-tute for natural TFIIIB activity in effecting transcription ofconventional class III genes (tRNA, VA1, and 5S RNA) withinternal promoter elements. The natural binary complex is alsohighly stable and remains intact through ion-exchange chro-matography, immunoprecipitation, and gradient centrifuga-tion at high salt (0.5 M ammonium sulfate) (data not shown).This stability is attributed to a strong direct interaction be-tween TBP and TFIIIB90, as verified by protein blot analysisof natural and recombinant proteins with a TBP probe.An important question is whether the TFIIIB90/TBP com-

plex represents the complete TFIIIB for transcription ofconventional class III genes, especially in light of evidence forheterogeneity in complexes containing TFIIIB90 and TBP(below) and indications from yeast (6) of a third TFIIIBcomponent (albeit one which is readily separated from theother components; ref. 9). Although transcription in a humansystem reconstituted with TFIIIB90/TBP, pol III, and highlypurified TFIIIC1 and TFIIIC2 (2, 19) requires another com-ponent present in both crude TFIIIB and TFIIIC fractions(data not shown), it is not clear whether this should beconsidered a part of natural TFIIIB.

Relationship of Minimal TFIIIB to Other TBP-ContainingComplexes. Previous studies (34, 35) have resolved two TBP-

Biochemistry: Wang and Roeder

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

7030 Biochemistry: Wang and Roeder

containing complexes from the conventional TFIIIB fraction(0.3 M KCl eluate from phosphocellulose). The more abun-dant B-TFIID is composed of TBP and a 180-kDa polypeptide(TAF180) and fails to manifest TFIIIB activity, whereas theother fraction, containing TFIIIB activity, is undefined inoverall polypeptide composition. Using ion-exchange and im-munoaffinity methods to purify TBP-containing complexesfrom a cell line stably transfected with an epitope (FLAG)-tagged TBP (17), we also have identified several componentsin the 0.3 M KCl eluate from phosphocellulose: (i) a fractionwith TFIIIB activity containing TBP, TFIIIB90, and a 67-kDapolypeptide of unknown function; (ii) a TFIID complex related(by TAF analysis) to the TFIID in the 0.85 M KCl phosphocel-lulose fraction; and (iii) the previously reported B-TFIID con-taining TBP and TAF180 (unpublished observations). In light ofthe present and previous results we believe that the TAF172reported by Taggart et al. (18) as a component of TFIIIB actuallyreflects contamination of the latter with the more abundantB-TFIID complex. Other pol III transcription activities detectedin the 0.3 M KCI phosphocellulose (TFIIIB) fraction include a"TFIIIB" component (TAF-L) weakly associated with TBP-containing complexes (18) and a "TFIIIB" activity (0.48 M KClMono Q fraction) separated from TBP-containing complexesupon Mono Q chromatography (16). Based on our analysis of theelution position of TFIIIB (0.28 M KCl) and pol III (0.35 M KCl)on a Mono Q column, we suspect that these activities representeither pol III or a factor associated in our preparation with pol III.The 87-kDa polypeptide in the TBP-containing 0.38 M KCIMono Q TFIIIB fraction of Chiang et al. (17) is the same as theTFIIIB90 reported here.Human TFIIIB90 Contains Functional TFIIB- and HMG2-

Related Domains. Human TFIIIB90 appears to be the func-tional homologue of yeast TFIIIB70, since it contains aTFIIB-related N-terminal domain which (like TFIIB) interactsweakly with TBP and a distinct C-terminal domain whichinteracts strongly with TBP. On the basis of the knownfunctions of yeast TFIIIB and human TFIIB, including inter-actions with corresponding RNA polymerases (10, 11, 25), it isexpected that human TFIIIB90 will have a similar role in polIII recruitment through concomitant interactions both withpol III and with TBP and promoter-bound TFIIIC. YeastTFIIIB70 has been shown to interact with specific subunits ofpol III (10, 11) and TFIIIC (8, 11) as well as with TBP. Thetiming and function of the N- and C-terminal interactions ofhuman TFIIIB90 with TBP remain to be determined and couldvary during preinitiation complex formation (6). Both regionsof human TFIIIB90 appear to be essential for function (Fig. 5),consistent with studies of yeast TFIIIB70 (14).

Interestingly, the C terminus of human TFIIIB90 differsconsiderably from that of yeast TFIIIB70 and contains se-quences related to those in HMG2 and in histone Hi-I (Fig.2B), both of which can interact nonspecifically with DNA.Along with the known DNA-binding properties of HMGdomains (28-30), these observations may provide some cluesas to the nature of the highly stable interaction of TFIIIB withthe promoter during preinitiation complex formation (4). Onespeculation is that an interaction with promoter-bound TFIIICmay promote conformational changes in TFIIIB, includingaltered TBP interactions, that lead to stable DNA interactions.

Role of Human TFIIIB90 in U6 and 7SK Transcription.Whereas past studies have implicated TBP in U6 and 7SKtranscription (for review, see ref. 2), the present study alsoindicates TFIIIB90 in this process. However, the inability ofnatural TFIIIB or a TBP/TFIIIB90 complex to restore 7SKand U6 transcription to a TFIIIB90-depleted nuclear extract,in contrast to the results observed for tRNA and VA1 RNApromoters, suggests the involvement of a TFIIIB90-containingcomplex distinct from that involved in 5S, tRNA, and VAl

transcription. This is in agreement with a previous study whichalso indicated that 7SK and U6 transcription requires TFIIIC1but not TFIIIC2, whereas 5S, tRNA, and VA RNA transcrip-tion requires both (2). Hence, the mammalian class III geneswith external proximal sequence and TATA promoter ele-ments (7SK and U6) seem different from those (5S, tRNA, andVA RNA) with internal promoters with regard to "general"factor requirements.

We thank A. Hoffmann for helpful discussions and for anti-TBPpolyclonal antibody, B. Moorefield for highly purified human TFIIIA,C. M. Chiang for the 0.3 M KCI phosphocellulose fraction from the cellline expressing FLAG-tagged TBP, and F. Brunel for protein sequencealignments. We thank Owen Witte for the human K562 cell cDNAlibrary and Joseph Schlessinger for the human 293 embryonic kidneycell cDNA library. Protein sequence analysis was provided by theRockefeller University Protein Sequencing Facility, which is sup-ported in part by National Institutes of Health shared instrumentationgrants and funds provided by the U.S. Army and Navy for purchase ofequipment. This work was supported by grants from the NationalInstitutes of Health (CA42567) to R.G.R. and from the Pew CharitableTrusts to Rockefeller University.

1.

2.

3.4.

5.6.

7.

8.

9.

10.

11.12.

13.

14.15.16.

17.

18.19.

20.

21.

22.23.

24.25.

26.27.

28.

29.

30.31.

32.33.34.35.

Geiduschek, E. P. & Kassavetis, G.A. (1992) Transcription Regulation(Cold spring Harbor Lab. Press, Plainview, NY), pp. 247-279.Yoon, J., Murphy, S., Bai, L., Wang, Z. & Roeder, R. G. (1995) Mo. Cell.Biol. 15, 2019-2027.Lassar, A. B., Martin, P. L. & Roeder, R. G. (1983) Science 222, 740-748.Kassavetis, G. A., Braun, B. R., Nguyen, L. H. & Geiduschek, E. P. (1990)Cell 60, 235-245.Struhl, K. (1994) Science 263, 1103-1104.Kassavetis, G. A., Joazeiro, C. A. P., Pisano, M., Geiduschek, E. P., Col-bert, T., Hahn, S. & Blanco, J. A. (1992) Cell 71, 1055-1064.Kassavetis, G. A., Bartholomew, B., Blano, J. A., Johnson, T. E. & Gei-duschek, E. P. (1991) Proc. Natl. Acad. Sci. USA 88, 7308-7312.Bartholomew, B., Kassavetis, G. A. & Geiduschek, E. P. (1991) Mol. Cell.Biol. 11, 5181-5189.Huet, J., Conesa, C., Manaud, N., Chaussivert, N. & Sentenac, A. (1994)Nucleic Acids Res. 22, 3433-3439.Werner, M., Chaussivert, N., Willis, I. M. & Sentenac, A. (1993) J. Biol.Chem. 268, 20721-20724.Khoo, B., Brophy, B. & Jackson, S. P. (1994) Genes Dev. 8, 2879-2890.Buratowski, B., Durkovich, D., Kassavetis, G. A. & Geiduschek, E. P.(1993) Mol. Cell. Biol. 13, 942-952.Lopez-De-Leon, A., Librizzi, M., Puglia, K. & Willis, I. M. (1992) Cell 71,211-220.Colbert, T. & Hahn, S. (1992) Genes Dev. 6, 1940-1949.Robert, S. G. & Green, M. R. (1994) Nature (London) 371, 717-720.Lobo, S. M., Lister, J., Sullivan, M. L. & Hernandez, N. (1991) Genes Dev.5, 1477-1489.Chiang, C., Ge, H., Wang, Z., Hoffmann, A. & Roeder, R. G. (1993)EMBOJ. 12, 2749-2762.Taggart, A. K. P., Fisher, T. S. & Pugh, B. F. (1992) Cell 71, 1015-1028.Sinn, E., Wang, Z., Kovelman, R. & Roeder, R. G. (1995) Genes Dev. 9,675-685.Dignam, J. D., Lebovitz, R. M. & Roeder, R. G. (1983) Nucleic Acids Res.11, 1475-1489.Segall, J., Matsui, T. & Roeder, R. G. (1980) J. Biol. Chem. 255, 11986-11991.Hoffmann, A. & Roeder, R. G. (1991) Nucleic Acids Res. 19, 6337-6338.Takada, R., Nakatani, Y., Hoffmann, A., Kokabo, T., Hasegawa, S.,Roeder, R. G. & Horikoshi, M. (1992) Proc. Natl. Acad. Sci. USA 89,11809-11813.Kovelman, R. & Roeder, R. G. (1992) J. Biol. Chem. 267, 24446-24456.Ha, I., Roberts, S., Maldonado, E., Sun,X.,Kim,L.,Green,M. & Reinberg,D. (1993) Genes Dev. 7, 1021-1032.Sparrow, D. B. & Wells, J. R. E. (1992) Gene 114, 289-290.Jantzen, H. M., Admon, A., Bell, S. P. & Tjian, R. (1990) Nature (London)344, 830-836.Weir, H. M., Kraulis, P. J., Hill, C. S., Raine, A. R. C., Lawe, E. D. &Thomas, J. D. (1993) EMBO J. 12, 1311-1319.Read, C. M., Cary, P. D., Robinson, C. C., Driscoll, P. C. & Norman, D. G.(1993) Nucleic Acids Res. 21, 3427-3436.Grosschedl, R., Giese, K. & Pagel, J. (1994) Trends Genet. 10, 94-100.Simmen, K. A., Bernues, J., Parry, K. D., Stunnenberg, H. G., Berkenstam,A., Cavallini, B., Egly, J. M. & Mattaj, I. W. (1991) EMBOJ. 10, 1853-1862.Moorefield, B. & Roeder, R. G. (1994) J. Biol. Chem. 269, 20857-20865.Santos, T. & Zasloff, M. (1981) Cell 23, 699-709.Timmers, H. T. M. & Sharp, P. A. (1991) Genes Dev. 5, 1946-1956.Meyers, R. E. & Sharp, P. A. (1993) Mol. Cell. Biol. 13, 7953-7960.

Proc. Natl. Acad. Sci. USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0