Protease Footprinting Analysis of Ternary Complex Formation by Human TFIIA*

(Received for publication, July 16, 1996, and in revised form, October 25, 1996)

Roderick Hori and Michael Carey Dagger

From the Department of Biological Chemistry, University of California, Los Angeles, California 90095-1737

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
Acknowledgments
REFERENCES


ABSTRACT

Transcription factor (TF) IIA performs two important regulatory functions during RNA polymerase II transcription: it is required for efficient binding of TFIID to a core promoter and it mediates the effects of upstream activators, both through direct interaction with the TATA box binding protein (TBP). To begin studying how TFIIA mediates these effects, we used a highly sensitive protease footprinting methodology to identify surfaces of human TFIIA participating in TFIIA·TBP·TATA ternary complex formation. Chymotrypsin and proteinase K cleavage patterns of TFIIA bearing a 32P-end-labeled gamma  subunit revealed that amino acids 59-73 were protected from cleavage both in the context of an immobilized ternary complex and in a binary complex with TBP alone. In contrast, amino acids 341-367 in the beta  portion of a 32P-labeled alpha -beta subunit were protected in the ternary but not in the binary complex, implying that those residues interact with promoter DNA. The regions of human TFIIA identified by protease footprinting are homologous to and encompass the yeast TFIIA residues that contact TBP and DNA in the recently solved crystal structure of the yeast ternary complex. The conservation of the regions and residues mediating complex formation implies that yeast and human TFIIA employ the same mechanism to stabilize the binding of TFIID to a core promoter.


INTRODUCTION

Transcription of protein-encoding genes by RNA polymerase II (pol II)1 is regulated by an intricate array of protein-protein and protein-DNA interactions (1, 2, 3, 4). Understanding how these interactions mediate formation of a transcription complex over a core promoter is a central problem in the field of gene expression. In the step-wise model, transcription complex assembly is nucleated by binding of the general initiation factors (<UNL>T</UNL>ranscription <UNL>F</UNL>actor) TFIID and TFIIA to the TATA box generating the "DA complex" (5). The parallel between the ability of gene activators to facilitate DA complex formation and to activate transcription suggests that the complex plays a key role in regulation (6).

TFIID is a multisubunit complex consisting of TATA box binding protein (TBP) and eight or more TBP-associated factors (TAFs) (7, 8, 9, 10). TBP alone, when it is substituted for TFIID, can nucleate the formation of a basal transcription complex that is fully functional but not responsive to activators. In contrast, TAFs are thought to act as co-activators because they are dispensable for basal transcription but are required to obtain activator-responsive transcription in vitro. In addition to TAFs, a fraction called upstream-factor stimulatory activity (USA), which contains both positive and negative co-regulators, potentiates transcriptional activation (11).

In the DA complex, TFIIA functions both to stabilize the relatively weak binding between TFIID and the TATA box and is necessary for activator recruitment of TFIID (12, 13, 14); its role in recruitment depends on the TAFs since activators have no effect on TBP·TFIIA complex formation. In the stepwise model, the formation of the transcription complex is completed by the successive association of TFIIB, RNA polymerase/TFIIF, TFIIE, and TFIIH (2, 4). An alternative model for complex formation involves a holoenzyme (15) containing RNA polymerase II and many of the other general initiation factors. The holoenzyme was discovered initially in yeast and more recently in mammalian cells. Both the yeast and some mammalian versions have been reported to lack TFIID and TFIIA (16, 17, 18). One possible function of the DA subcomplex is to form an activator-responsive platform for recruitment of the holoenzyme.

Human (and Drosophila) TFIIA is a multisubunit protein consisting of three subunits called alpha  (LN), beta  (LC), and gamma  (S) (19, 20, 21, 22, 23, 24, 25). alpha  and beta  are synthesized as a precursor that is processed proteolytically to generate the mature subunits although the unprocessed form is functional in vitro. Yeast TFIIA consists of only two subunits, TOA1 and TOA2 (26, 27). TOA1 is homologous at its amino and carboxyl termini to regions of the human alpha  and beta  subunits, respectively, and TOA2 is homologous to the human gamma  subunit (19, 20, 21, 22). Analysis of systematic internal deletion mutants of both subunits demonstrated that all of the regions conserved between yeast and human TFIIA were required for yeast viability (28). For TOA1 (alpha  and beta ), these deletion mutants defined the amino- and carboxyl-terminal ends as essential but the nonconserved, middle region as dispensable for viability. A deletion mutant removing residues 217-240 abolishes binding to TBP. For TOA2 (gamma ), all of the internal deletions but one resulted in inviable yeast, making it difficult to genetically define functional domains. Alanine scanning mutants of both subunits were screened for temperature-sensitive (ts) growth phenotypes. In TOA1, all the ts mutations were in basic residues between residues 253 and 259. These mutants bound normally to TBP but could not form complexes, implying a decreased ability to bind DNA. Among the three ts mutants identified in TOA2, one double mutant, D73A/D74A, bound less tightly to TBP although complex formation was unaffected, and the other two exhibited no detectable difference in their ability to bind TBP or form ternary complexes. Human TFIIAgamma has also been analyzed by alanine substitution mutations and Y65A (equals Y69 in yeast TOA2) caused the greatest decrease in the ability of IIA to stabilize TBP binding (29).

The crystal structure of the ternary complex containing the yeast (y)CYC1 TATA box, yTBP, and yTFIIA was recently published (30, 31). In the crystal structure, yTBP, which encodes two 80-amino acid direct repeats, folds into a pseudosymmetric saddle-like shape containing a hydrophobic concave surface composed of a 10-strand antiparallel beta -sheet, which interacts with the minor groove of the DNA. TBP binding distorts the TATA box by inserting phenylalanine residues between the first and final base pairs of the recognition site, resulting in bending of the DNA by 80° and broadening of the minor groove (32, 33). yTFIIA, on the other hand, is composed of a four-helix bundle and a 12-strand beta -barrel (30, 31). The beta -barrel is made of six strands each from the carboxyl terminus of TOA1 (TOA1C~human TFIIAbeta ) and the carboxyl terminus of TOA2 (gamma ). The four-helix bundle is composed of the amino terminus of TOA1 (TOA1N~human TFIIAalpha ) and the amino terminus of TOA2. The interaction between TFIIA and TBP places the beta -sheets of TOA2 and TBP in close proximity, resulting in a continuous 16-strand beta -sheet. TFIIA interacts with DNA through basic residues in TOA1C that lie within a loop connecting two beta -strands.

DNA-protein cross-linking studies have confirmed certain aspects of the structure in solution (34, 35). Collectively, these studies show cross-linking was observed between DNA and alpha  upstream of the TATA box, between beta  and both sides of the TATA box, and with gamma  at one position upstream of the TATA. In the crystal structure of the yeast complex, only a region homologous to part of the beta  subunit contacts DNA.

To study how TFIIA stabilizes TFIID/TBP binding, we investigated the assembly of the ternary complex containing a mammalian TATA box, human TBP, and human TFIIA. Presently, there is no genetic analysis of the alpha -beta subunit of human TFIIA or crystal structure of free TFIIA. To identify the interactions mediating complex formation, the differences between free TFIIA and complexed TFIIA were characterized using protease footprinting. The interactions of TFIIA with TBP and DNA were identified along with the accessible regions of TFIIA. The results of our protease footprinting studies are placed within the context of the recently solved x-ray structure of the yeast ternary complex.


EXPERIMENTAL PROCEDURES

DNA Constructs

To engineer vectors containing both His and HMK tags at alternate termini, pET21a-N-HMK was constructed by cloning an oligonucleotide, 5'-TATGCGCCGCGCCAGTGTGGGATCCC-3', encoding the HMK phosphorylation site (HMK tag) between the NdeI and XhoI sites of pET21a (Novagen). pET11d-N-His/C-HMK was constructed by cloning an oligonucleotide, 5'-CTAGCCACCACCACCACCACCACG-3', encoding the His tag between the NheI and BamHI sites of pET11d-C-HMK (36). A BamHI fragment encoding TFIIA alpha -beta and a BamHI-BglII fragment encoding TFIIA gamma coding regions were synthesized by PCR and ligated into the BamHI site of pET11d-N-His/C-HMK and pET21a-N-HMK, respectively, using T4 ligase. pN-HMK-TFIIAgamma Delta 81 was made by digesting pN-HMK-TFIIAgamma -C-His at the EcoRI site within the coding region, repairing with the Klenow fragment and deoxynucleoside triphosphates, and inserting a stop codon oligonucleotide, 5'-CTAATCTAGATTAG-3', using T4 DNA ligase. pC-HMK-TFIIAalpha -beta Delta 38, Delta 129, Delta 254 were constructed by deleting the coding region between the NheI and ScaI (Delta 38), NdeI (Delta 129), and MscI (Delta 254) sites within the coding regions, respectively, by digesting with these restriction enzymes, filling in with the Klenow fragment, and ligating using T4 DNA ligase. These constructs encode internal deletions between residues 3 and 38, 129 and 254, respectively.

Protein Purification

TFIIA and GST-TBP (The GST-TBP expression vector was a generous gift of Arnie Berk, UCLA) were purified as described (20, 37). TFIIA containing the HMK tag on the carboxyl terminus of the alpha -beta subunit was further purified by affinity chromatography over a GST-TBP affinity resin (28). The TFIIA deletion constructs were transformed into BL21(DE3) and inoculated into 250 ml of Luria broth containing 50 µg/ml ampicillin. When the cell culture reached an A600 of 0.6, protein expression was induced for 1.5 h by the addition of isopropyl-1-thio-beta -D-galactopyranoside to 0.3 mM. All of the following steps were performed at 4 °C. The cells were collected by centrifugation in a Sorvall SA600 rotor at 5,000 rpm for 5 min, washed with Buffer A (20 mM HEPES, pH 7.9, 0.2 mM EDTA, 0.5 M KCl), collected again by centrifugation at 5,000 rpm for 5 min, resuspended in 50 ml of Buffer A containing 0.5 mM DTT, 0.5 mM PMSF, 50 µg/ml benzamidine, 1 µg/ml leupeptin, and 1 µg/ml pepstatin A, and lysed by sonication. For TFIIAgamma Delta 81 and for TFIIAalpha -beta Delta 38 and Delta 129, the membrane fraction was extracted with 30 ml of Buffer D (20 mM HEPES, pH 7.9, 0.2 mM EDTA, 20% glycerol) containing 0.2 M KCl, 8 M urea, 0.5 mM DTT, and 0.5 mM PMSF. The urea was removed by step-wise dialysis into Buffer D containing 0.3 M KCl, 0.5 mM DTT, and 0.5 mM PMSF and by decreasing concentration of urea from 2 to 0.5 to 0 M. For TFIIAalpha -beta Delta 254, the lysate was precipitated by successive additions of 0.20, 0.25, and 0.30 g/ml of (NH4)2SO4, stirring for 30 min, and then collecting the precipitate by centrifugation at 10,000 rpm for 10 min. TFIIAalpha -beta Delta 254 was in the 0.25 g/ml precipitate and was solubilized in 3 ml of Buffer D containing 0.3 M KCl, 0.5 mM DTT, and 0.5 mM PMSF. Recombinant human TBP was a generous gift from Zong Juo and Richard Dickerson (UCLA) (38).

Complex Formation and Proteolysis

TBP-TFIIA-TATA box ternary complexes were formed in a 50-µl mixture containing 3 pmol of biotinylated adenovirus E4 DNA template, a synthetic biotinylated oligonucleotide (biotin-5'-GGATCCCCAGTCCTATATATACTCGCTCTGC-3') immobilized on streptavidin-conjugated M-280 magnetic beads (Dynal), with 500 ng of TBP, and 400 ng of 32P-labeled TFIIA. After 30 min at 30 °C, the complexes were washed 3 times with 100 µl of binding buffer (12 mM HEPES, pH 7.9, 12% glycerol, 60 mM KCl, 2 mM MgCl2, 0.02% Nonidet P-40, 0.5 mM DTT) and digested with 5-200 ng of either proteinase K (Promega) or chymotrypsin (Sigma) alongside free TFIIA or TFIIA incubated with immobilized DNA. TFIIA incubated with immobilized DNA was only washed once to maintain some of the protein in the presence of the resin for the control reaction. After 1.5 min, the cleavage reactions were terminated by addition of PMSF to 10 mM and freezing in dry ice. SDS-loading dye was added, and the cleavage products were fractionated on 15% total acrylamide-3.3% cross-linker or 16.3% total acrylamide/5.0% cross-linker tricine-SDS gels. Binary complexes were formed and footprinted as described (36). Protease sequencing ladders were generated by denaturing 120 ng of labeled TFIIA in 10 mM HEPES, pH 7.9, 10% glycerol, 50 mM KCl, 2 M urea, 1 mM DTT, and 0.1% SDS at 50 °C and digesting at 37 °C with 3 µg of endoproteinase Lys-C (1-10 min), 1.5 µg of clostripain (5-90 min), or 5 µg endoproteinase Glu-C (1-10 min) (Promega). The cleavage sites were assigned using a combination of the known amino acid sequence, prestained molecular weight markers (Bio-Rad), and deletion markers. The positions of the digestion products in the experimental lanes were mapped using all of these markers and a semi-log plot of mobility based on these markers. The precision of the semi-log plot in the protected regions is ±2 residues. The gels were autoradiographed by exposure to XAR-5 film. The autoradiographs were scanned into Adobe Photoshop 3.0 using a ScanMakerIII (Microtek), imported into Microsoft PowerPoint 4.0, and labeled and printed onto glossy paper on a Tektronix printer. For quantitation, either the autoradiographs were scanned with a laser densitometer or the gel was exposed to a PhosphorImager screen and quantitated using ImageQuant software (Molecular Dynamics). Sample lanes in which the extent of digestion was similar, as judged by the percentage of full-length material remaining in the experimental and control lanes, were compared, and the ratio of these signals was used to normalize other signals. To determine the extent to which each band was protected, the signal ratio of a digestion product in the experimental lane to that of the control lane was calculated and then normalized using the signal ratio of full-length protein.


RESULTS

Protease Footprinting Methodology

Protease footprinting represents a potentially powerful technology for examining protein-protein and protein-DNA interactions. In its simplest form, the concept is analogous to DNase I footprinting except that, rather than employing nucleases to map a protein binding site on a 32P-end-labeled DNA fragment, proteases are used to map protein contacts on a 32P-end-labeled target protein. The advantage of protease footprinting over deletion and point-mutation analyses is that interactions are studied in the context of a largely native wild-type protein, bypassing the potential for deleterious effects that mutants might have on the overall structure. Furthermore, because the complexed protein is compared with the free protein, conformational changes upon interaction may also be detected quite readily.

In a previous study, we employed protease footprinting to map the binding sites of the viral activator VP16 and Drosophila TAFII40 on human TFIIB (36). The VP16 binding sites correlated with those defined by mutational studies (39), whereas the TAF40 binding site was not previously identified. The subsequent crystal structure of the TBP-TFIIB-TATA box ternary complex (40) revealed that the VP16 and the TAF40 binding sites defined by protease footprinting formed a solvent-accessible wedge and were in close proximity, consistent with their ability to mediate complex formation (41).

Our current study characterizes the interactions of human TFIIA within the TFIIA-TBP-TATA ternary complex to understand how TFIIA assists TBP and TFIID binding to DNA, a step that serves as a key target for transcriptional activators. Although our study was performed independently, comparison with the recently solved crystal structure of the yeast ternary complex validates the utility of the protease footprinting technology for studying intricate protein-protein interactions.

Experimental Design

Fig. 1 describes the protease footprinting procedure used in our study. The TFIIA used for these experiments contains a unique 32P-end label on one of its two subunits (alpha -beta or gamma ) (Fig. 1A). The 32P-end-labeled TFIIA, the free molecule, or that in the form of either a binary complex with GST-TBP or a ternary complex with TBP bound to an immobilized adenovirus E4 TATA box oligonucleotide is digested with a nonspecific protease to generate a nested set of digestion products (Fig. 1B). The products are fractionated on a high resolution SDS-polyacrylamide gel and autoradiographed to reveal a ladder (Fig. 1B). The limited proteolysis pattern of TFIIA in the complex is compared with that of free TFIIA to identify changes in cleavage that occur as a consequence of complex formation. The loss (or increase) of specific cleavage products is taken as an indicator of an interaction. To determine whether these footprints are due to a direct interaction with TBP or the TATA box, protease footprinting of GST-TBP·TFIIA binary complexes formed in the absence of DNA is compared with the results obtained with the ternary complexes.


Fig. 1. Experimental design. A, specific 32P-end labeling of TFIIA. Recombinant TFIIA subunits bearing a heart muscle kinase phosphorylation site (HMK tag) at one terminus and a His-tag at the other were overexpressed and purified from Escherichia coli by Ni-NTA chromatography. The tagged subunits were then joined with a partner lacking the HMK tag and labeled with [gamma -32P]ATP and heart muscle kinase to generate TFIIA 32P-end-labeled at the HMK tag of one subunit. B, schematic of protease footprinting protocol. To form ternary complexes, recombinant human TBP (the carboxyl-terminal 181 amino acids) and human 32P-end-labeled TFIIA were incubated with a biotinylated oligonucleotide containing the TATA box from the adenovirus E4 promoter immobilized on streptavidin:magnetic beads. (Binary complexes were assembled by incubating glutathione-agarose-immobilized GST-TBP with human 32P-end-labeled TFIIA, not shown.) The complexes were washed with an excess of binding buffer to remove unbound protein and then digested with limiting amounts of a broad specificity protease. The products were fractionated on high resolution SDS-polyacrylamide gels and visualized by autoradiography. C, specific ternary complex formation. Increasing amounts of TFIIA end-labeled on the gamma  subunit were incubated with biotinylated TATA box oligonucleotides immobilized on M-280 streptavidin resin with (lanes 3, 5, 7, and 9) or without (lanes 2, 4, 6, and 8) TBP and then washed extensively. The bound fraction was electrophoresed on standard SDS-polyacrylamide gels and autoradiographed. Lane 1 is 3% of the input end-labeled TFIIA for the experiments shown in lanes 8 and 9.
[View Larger Version of this Image (35K GIF file)]


32P-end-labeled TFIIA was generated by constructing molecules of TFIIA in which one subunit had an HMK tag (Arg-Arg-Ala-Ser-Val; RRASV) engineered onto one of its termini and a His tag onto the other for purification using nickel affinity resins (see Fig. 1A). These molecules were phosphorylated on the HMK tag using [gamma -32P]ATP and heart muscle kinase; the labeling was specific as equimolar amounts of a TFIIA molecule containing untagged subunits incorporated less than 5% of the amount of 32P generated using tagged subunits (data not shown).

Fig. 1C shows an example of how the binding reactions were optimized. Increasing amounts of TFIIA, 32P-end-labeled on the gamma  subunit, were first incubated with a resin containing an immobilized TATA box and subsequently washed to remove unbound protein. The bound fraction was electrophoresed on an SDS-polyacrylamide gel and autoradiographed. Fig. 1C illustrates that specific ternary complexes were obtained based on the TBP dependence of TFIIA binding (compare even-numbered lanes with odd-numbered lanes). At the highest concentration point in this experiment, 34% of the input TFIIA labeled on the gamma  subunit was assembled into ternary complexes. The HMK-tagged TFIIA molecules used here have been shown to also be competent to form TATA box-TBP-TFIIA complexes in electrophoretic mobility shift assays (data not shown).

The TFIIAgamma Subunit Binds Directly to TBP

Fig. 2, A and B, reveals a strong interaction between a small region of the TFIIAgamma subunit and TBP, which apparently tethers TFIIA to the ternary complex. Fig. 2A shows chymotrypsin and proteinase K footprints of a ternary complex containing TFIIA 32P-labeled on the amino terminus of the gamma  subunit. The chymotrypsin digestion patterns of free TFIIA (lanes 2-5) or TFIIA in the presence of only the E4 TATA box oligonucleotide (lanes 11-15) were identical except for a cleavage site that is enhanced in the presence of promoter DNA (asterisk). In contrast, the cleavage pattern of TFIIA assembled into a ternary complex (lanes 6-10) revealed a footprint between amino acid residues 51 and 73 as measured by the Lys-C and clostripain protease sequencing ladders (lanes 28-29, respectively). The position of the Lys-C and clostripain cleavages were assigned by mapping the known amino acid sequence and prestained markers against the migration of the digestion products on a semi-log plot. Proteinase K footprinting revealed that a subset of the same residues (amino acids 59-73) was protected in the ternary complex (compare lanes 20-23 with lanes 16-19 and 24-27) with concomitant DNA-dependent enhanced cleavages (asterisks) flanking the protected region. Using laser densitometry, the three cleavage products between residues 59 and 73 in the proteinase K footprint were 79%, 93%, and 79% protected in the complex versus the free TFIIA (see "Experimental Procedures" for normalization procedure).


Fig. 2. Protease footprinting of the gamma  subunit. A, ternary complexes. Ternary complexes containing the E4T core promoter, TBP, and TFIIA 32P-labeled on the amino terminus of the gamma  subunit (E4T-TBP-IIA) were assembled and subjected to proteolysis by increasing concentrations (filled triangles) of chymotrypsin or proteinase K (lanes 7-10 and 20-23, respectively) in parallel with free TFIIA (lanes 3-5 and 16-19) and TFIIA incubated in the presence of the TATA box oligonucleotide alone (E4T-IIA, lanes 12-15, and 24-27). Mock digestions (0) to identify background proteolysis are shown in lanes 2, 6, and 11. Lane 1 contains 0.2% of the input TFIIA used in the binding reactions. The bracket on the left marks the positions of the protected sites, and asterisks indicate TATA oligo-enhanced cleavage sites. In the case of proteinase K, these sites can be detected without DNA when there is a slightly more extensive digest. Lanes 28 and 29 show sequencing ladders generated by Lys-C and clostripain (cuts at Arg). The Lys and Arg cleavages can be aligned against the primary sequence to generate a ladder of markers whose positions are indicated on the right, adjacent to a schematic of TFIIAgamma . "Bgl" represents the position of an arginine residue encoded by the BglII site used for cloning. The HMK tag and His tag are represented by dark and light cross-hatching, respectively. The positions of the first and last wild-type residues are denoted by 1 and 109, and an asterisk near the bottom represents the position of the 32P-end-label. B, binary complexes. GST-TBP·IIA binary complexes were assembled and subjected to proteolysis by proteinase K (lanes 7-9) in parallel with free TFIIA (lanes 11-13) and TFIIA in the presence of GST alone as a control (lanes 3-5). Mock digestions to identify any background proteolysis are shown in lanes 2, 6, and 10. Lane 1 contains 0.2% of the input TFIIA used in the binding reactions. A bracket on the left identifies the protected region. Lys-C and clostripain sequencing ladders are shown in lanes 14 and 15. Lane 16 contains a carboxyl-terminal deletion of TFIIAgamma to residue 81. The estimated positions of the lysines and arginines are shown on the right. TFIIAgamma appears to interact with TBP at largely the same positions as in the TBP-TATA complex. In the binary complex, amino acids 59-73 of human TFIIAgamma , which correspond to amino acids 63-77 in Saccharomyces cerevisiae TOA2, are protected.
[View Larger Version of this Image (70K GIF file)]


Protease footprinting of the binary TBP-TFIIA complex (Fig. 2B) demonstrates that the protection described above represents a direct interaction with TBP. When compared with free TFIIA (lanes 10-13) or TFIIA incubated with GST alone (lanes 2-5), proteinase K digestion of a binary complex between TFIIA and GST-TBP immobilized on glutathione agarose (lanes 6-9) revealed a footprint. Alignment against the sequencing ladder and deletion marker (lanes 14-16) revealed that the footprint encompassed the same region (residues 59-73) of the TFIIAgamma subunit as that identified in the ternary complex (Fig. 2A) with the minor difference that the DNA-dependent enhanced bands were absent. When samples digested to a comparable extent were quantitated using a PhosphorImager and compared, there was >90% inhibition of proteolysis of the lowest three cleavage products on the gel. Taken together, our results suggest that residues 59-73 of the gamma  subunit mediate a direct protein-protein interaction with TBP. The residues in human TFIIAgamma identified by protease footprinting, 59-73, are homologous to residues 63-77 in yeast TOA2, which completely encompasses the TBP-interacting amino acid residues in the crystal structure. We will return to this point in the discussion (30, 31). The digestion pattern also reveals that the gamma  subunit is divided into a protease-resistant amino-terminal half, which precludes a protease footprinting analysis of this region, and a more protease-sensitive carboxyl-terminal half (see Figs. 2A and 2B).

The TFIIAalpha -beta Subunit Makes a Direct Contact with DNA via the beta  Subunit

Fig. 3A demonstrates that TFIIA makes a direct contact with DNA via the beta  portion of the alpha -beta subunit. TFIIA containing alpha -beta 32P-labeled at the carboxyl terminus was assembled into a ternary complex and footprinted using chymotrypsin. A striking feature of this digest is the very small number of accessible chymotrypsin cleavage sites, only 5 throughout the length of this subunit. Similar results were obtained using several proteases. The digestion patterns between free TFIIA (lanes 11-13) and TFIIA in the presence of the promoter alone (lanes 3-5), with the exception of background bands (see pound signs), are identical. However, when assembled into the ternary complex, the three sites nearest the carboxyl terminus become protected (lanes 7-9). The region identified, residues 341-367 (see Fig. 3A, brackets), extends from a basic region almost to the carboxyl terminus of the protein based on the clostripain, Lys-C and Glu-C cleavage sites (see Fig. 3A, lanes 14-16). Footprints in the same region were observed using alkaline protease and proteinase K (data not shown).


Fig. 3. Chymotrypsin footprinting of the alpha -beta subunit. A, ternary complexes containing TFIIA labeled on the carboxyl terminus of the alpha -beta subunit were assembled and subjected to proteolysis by chymotrypsin (lanes 7-9) in parallel with free TFIIA (lanes 11-13) or TFIIA in the presence of the TATA box oligo alone (lanes 3-5). Lane 1 contains 0.2% of the input TFIIA used in the binding reaction, and lanes 2, 6, and 10 are mock digestions. The brackets identify the protected regions, which include sites in the basic region of the beta  subunit (amino acids 341-354). The pound signs denote the presence of background proteolysis products. A schematic of TFIIAalpha -beta is on the right with the position of the basic region denoted by plus signs. The HMK tag and His tag are represented by dark and light cross-hatching, respectively. Lys-C, clostripain, and Glu-C sequencing ladders are shown in lanes 14-16, respectively. B, binary complexes. GST-TBP·IIA binary complexes were assembled and subjected to proteolysis by chymotrypsin (lanes 7-9) in parallel with free TFIIA (lanes 11-13) and TFIIA in the presence of GST alone as a control (lanes 3-5). Mock digestions to identify any background proteolysis are shown in lanes 2, 6, and 10. Lane 1 contains 0.2% of the input TFIIA used in the binding reactions. Asterisks on the left identify the regions protected in the ternary complex. A pound sign identifies a background proteolytic product that occurs in the presence of the GST resin. Deletion markers are in lanes 14-16, and clostripain, Lys-C, and Glu-C sequencing ladders are shown in lanes 17-19, respectively. The estimated positions of the lysine, arginine, and glutamic acid residues are shown on the right.
[View Larger Version of this Image (47K GIF file)]


The protection of the IIAalpha -beta observed in the ternary complex, unlike the case of the gamma  subunit, is not due to a direct interaction with TBP. This was illustrated by the lack of a footprint in the binary TBP-TFIIA complex (Fig. 3B); compare, for example, the digestion pattern when the binary complex is formed (lanes 6-9) with that of free TFIIA (lanes 10-13) and TFIIA in the presence of GST (lanes 2-5). Because no direct interaction between TBP and IIAalpha -beta can be detected, we conclude that the protection observed in the ternary complex is a consequence of an interaction between promoter DNA and TFIIAalpha -beta . This region contains residues Arg-344, Lys-346, Lys-348, and Lys-350, which are similar to residues Arg-253, Lys-255, Arg-257, and Lys-259 of yeast TOA1, that have been shown to contact DNA both in the crystal structure of the ternary complex (30, 31) and via a genetic analysis (28).


DISCUSSION

We have identified, using protease footprinting, interactions made by human TFIIA upon assembly of the TFIIA-TBP-TATA box ternary complex. Each subunit makes a series of contacts with one of the other components of the complex, the TFIIAgamma subunit with TBP and the IIAalpha -beta subunit with DNA. These two interactions allow TFIIA to form a bracket that stabilizes TBP binding. These results are summarized schematically in Fig. 4. Together, TBP and TFIIA function to provide specificity and stability. The preference of TBP for the TATA box provides the specificity required to bring TFIID to the promoter, but the weak affinity of TFIIA for DNA is used advantageously to stabilize binding of TFIID to promoters, which contain diverse sequences flanking the TATA box.


Fig. 4. Schematic summary of results. Schematics of the human TFIIA subunits are shown. The regions of TFIIAalpha -beta that are homologous with TOA1, the acidic region (- - -), the basic region (+), and the putative division between the alpha  and beta  subunits (vertical line at 254) are shown (20). The regions identified by protease footprinting to interact with TBP and DNA are represented by black boxes along with their respective residue numbers on the sides of the boxes (and in parentheses, the corresponding residue numbers in the yeast homologues). At the bottom of the figure, the regions identified in TOA1 to interact with DNA by Kang, et al. (28) are shown with a stippled box along with the residue numbers.
[View Larger Version of this Image (21K GIF file)]


Correlations with Crystal Structure and Biochemical Data

The contacts made by TFIIA within the human ternary complex as measured by protease footprinting are similar to those made within the yeast ternary complex as identified in the crystal structure. Fig. 5A is a ribbon diagram illustrating the recently solved crystal structure of the yeast ternary complex (30, 31) in which yTOA1 (alpha  and beta ), yTOA2 (gamma ), TBP, and the TATA box oligonucleotide are yellow, green, blue, and white, respectively. In the structure, the two subunits of yTFIIA are highly intertwined and form a boot-shaped structure in which TOA2 (gamma ) interacts with TBP and TOA1 (alpha  and beta ) interacts with DNA. TFIIA binds to the underside of TBP along the loop (stirrup) (42) connecting beta -strand 2 and beta -strand 3. A mutagenesis study, in which radical changes were introduced into surface residues of human TBP, identified an interaction between TFIIA and residues encompassing the stirrup within the first direct repeat of TBP, consistent with the crystal structure (43). The ability of these TBP mutants to support transcriptional activation in vivo and TBP-TFIIA complex formation in vitro was greatly diminished.


Fig. 5. Summary of results modeled on the crystal structure. A-D, ribbon model of the yTFIIA-yTBP-TATA box ternary complex. Yeast TFIIA (homologous to TFIIAalpha -beta ), yTOA2 (homologous to TFIIAgamma ), yTBP, and DNA containing a TATA box are yellow, green, light blue, and white, respectively. The view in panel A is from the same angle used in panels B and C. The view in panel D is the same one used in panels E and F. B, space-filling model of yTFIIA. Yeast TFIIA (homologous to TFIIAalpha -beta ) and yTOA2 (homologous to TFIIAgamma ) are yellow and green, respectively. The residues of hTFIIA (i.e. the homologous ones in yTOA2) identified by protease footprinting as interacting with TBP are identified as red. C, space-filling model of the ternary complex. yTFIIA is shown from the same view as in panel B and using the same color scheme. yTBP and DNA containing a TATA box are light blue and white, respectively. The interaction between the red residues of the gamma  subunit and TBP can be easily visualized. E, space-filling model of yTFIIA. The color scheme is the same as in panel B except that the residues of hTFIIA (i.e. the homologous ones in yTOA1) identified by protease footprinting as interacting with DNA are identified as red. F, space-filling model of the ternary complex. yTFIIA is shown from the same view as in panel E and using the same color scheme. yTBP and DNA containing a TATA box are light blue and white, respectively. The interaction between the red residues of the alpha -beta subunit and DNA can be easily visualized.
[View Larger Version of this Image (86K GIF file)]


In Figs. 5B, C, E, and F, our footprinting results are superimposed on space-filling models of yeast TFIIA (Fig. 5, B and E) alone and in the ternary complex (Fig. 5, C and F). Fig. 5, B and C, shows the gamma -TBP interaction using the same view as in Fig. 5A. Fig. 5B illustrates yTFIIA alone in which TOA1 (i.e. hTFIIAalpha -beta ) is yellow and TOA2 (i.e. hTFIIAgamma ) is green. The residues identified by protease footprinting as interacting with TBP (hTFIIAgamma residues 59-73 homologous to yTOA2 residues 63-77) are colored red. Of the 15 residues in the protected region, 11 of the positions are homologous between human and yeast including all those that directly contact TBP.

In Fig. 5C, TBP (blue) and the TATA box oligonucleotide (white) have been added to TFIIA. The protected region in red forms beta -strands that are part of the domain that is intertwined with the beta -strands of yTOA1 to form the beta -barrel. A comparison of Fig. 5B with 5C illustrates that the edge of the concave beta -sheet region of TBP covers the "red" region and explains why TBP protects that region from proteolysis.

Fig. 5, D-F, is a separate view of the ternary complex used to optimize presentation of the region of hTFIIAalpha -beta protected in the ternary complex. Fig. 5, E and F, superimposes the interaction between TFIIAalpha -beta and DNA identified by protease footprinting onto TFIIA alone (Fig. 5E) and in the ternary complex (Fig. 5F). In this case, the residues protected in hTFIIAalpha -beta (TOA2) are labeled as red, and this region lies within the beta -strands of the beta -portion of hTFIIAalpha -beta . Within the identified region, 16 of the 27 residues are homologous between yeast and human, including 6 basic ones.

The region of the human TFIIAalpha -beta subunit protected from proteolysis in the ternary complex includes the homologous residues of yeast TOA1 shown to contact DNA directly. In Fig. 5F, when yTBP (blue) and the TATA box oligonucleotide (white) are added to yTFIIA (shown from the same view and with the same color scheme in Fig. 5E), the DNA runs across the "red" residues, which is consistent with the observed protease footprint. It appears from the figure that the protected residue at the bottom of the figure may still be accessible from underneath. However, the TATA box oligonucleotide used for our footprinting experiments is longer than the one used in the crystallography studies, which would increase the protection of that site.

The one interaction observed in the crystal structure that was not detected in our study was the insertion of the penultimate tryptophan residue of yTOA1 into a crevice created where TBP and yTOA2 interact. Because that tryptophan is the carboxyl-terminal residue in hTFIIAalpha -beta , it is probably too close to our HMK tag for detection.

Although the protease footprinting analysis does not identify other interactions of hTFIIA within the ternary complex, it is possible that a protease-resistant region forms other contacts. The lack of a proteolytic reagent that cleaves at every surface position does tend to limit the protease footprinting approach. One possible limitation is revealed by mutational studies that indicate that an interaction exists between a highly acidic region of hTFIIA alpha -beta and the H2 helix of TBP (44, 45, 46). We have not identified such a region in our study.

Interactions of TFIIA with TBP in the absence of DNA have been studied using "GST-pulldown" experiments. Interactions of the alpha , beta , and gamma  subunits (or their homologues) individually with TBP have been identified. However, in view of the crystal structure, the previous studies will have to be reevaluated because the subunits are intertwined and probably will not fold properly on their own. Our studies define an interaction between only the gamma  subunit and TBP.

A protein-DNA cross-linking study concluded the alpha  subunit interacts with promoter DNA upstream of TBP centered at -45, the gamma  subunit is close to only -39, and the beta  subunit can be cross-linked to both sides of TBP (35). These data are similar to an earlier study (34). We detect only an interaction between the beta  portion of the alpha -beta subunit and DNA. The absence of an alpha -DNA interaction may simply reflect the use of a TATA box oligonucleotide that is too short to detect this interaction. It may also be possible that some proportion of TFIIA undergoes a conformational change within the ternary complex leading to these other DNA contacts, a finding supported by at least one recent study (21).

Based on protease sensitivity, it is plausible that other proteins will interact with TFIIA via the carboxyl terminus of the gamma  subunit and with the alpha  subunit. Because the gamma  subunit binds TBP, regulatory proteins interacting with the gamma  subunit could modulate the ability of TFIIA to associate with TBP. There is evidence that the Epstein-Barr virus activator ZEBRA overcomes a mutation in the gamma  subunit, which decreases TBP binding and complex formation (29). Because the alpha  subunit is upstream within the ternary complex (30, 31), alpha  is positioned to more easily interact with activators or possibly TAFs that extend toward upstream bound activators.

Our understanding of how the transcription complex is assembled requires an understanding of each step, particularly the key regulatory ones. We have characterized the formation of a promoter-bound complex containing TBP and TFIIA. We will use this study as a starting point to perform parallel studies with TFIID and to examine the interactions of activators. It will be particularly interesting to determine how (and if) TAFs change the interaction of IIA with TBP. The ability of activators to both facilitate complex assembly and induce a conformational change in the DA complex could be studied. These future studies will benefit from using smaller proteolytic reagents, especially given the larger size of TFIID relative to TBP. Recent experiments by the labs of Meares, Heyduk, and Ebright and coworkers (47, 48, 49) have shown that hydroxyl-radicals can be an effective reagent for protease footprinting studies and will probably be a useful reagent for our studies.


FOOTNOTES

*   This work was funded by National Institutes of Health Grant GM46424 (to M. C.) and by National Institutes of Health postdoctoral fellowship and Jonsson Cancer Center Foundation fellowship (to R. H.). The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dagger    To whom correspondence should be addressed: Dept. of Biological Chemistry, Center for Health Sciences 33-257, UCLA School of Medicine, 10883 LeConte Ave., Los Angeles, CA 90095-1737. Tel.: 310-206-7859; Fax: 310-206-9598; E-mail: mcarey{at}biochem.medsch.ucla.edu.
1    The abbreviations used are: pol II, RNA polymerase II; TF, transcription factor; TBP, TATA box binding protein; TAF, TBP-associated factor; USA, Upstream Factor Stimulatory Activity; ts, temperature-sensitive; HMK, heart muscle kinase; GST, glutathione-S-transferase; PMSF, phenylmethylsulfonyl fluoride, DTT, dithiothreitol.

Acknowledgments

We thank Tim Richmond for providing the coordinates of the TFIIA-containing ternary complex, Dimitar Nikolov and Steve Burley for providing the TFIIB-containing ternary complex, Richard Ebright and Arnie Berk for insightful discussions/comments on the manuscript, and Thai Nguyen for constructing some of the TFIIA deletion mutants.


REFERENCES

  1. Tjian, R., and Maniatis, T. (1994) Cell 77, 5-8 [Medline] [Order article via Infotrieve]
  2. Zawel, L., and Reinberg, D. (1995) Annu. Rev. Biochem. 64, 533-561 [CrossRef][Medline] [Order article via Infotrieve]
  3. Hori, R., and Carey, M. (1994) Curr. Opin. Genet. & Dev. 4, 236-244 [Medline] [Order article via Infotrieve]
  4. Conaway, R. C., and Conaway, J. W. (1993) Annu. Rev. Biochem. 62, 161-190 [CrossRef][Medline] [Order article via Infotrieve]
  5. Zawel, L., and Reinberg, D. (1993) Prog. Nucleic Acid Res. Mol. Biol. 44, 67-108 [Medline] [Order article via Infotrieve]
  6. Chi, T., Lieberman, P., Ellwood, K., and Carey, M. (1995) Nature 377, 254-257 [CrossRef][Medline] [Order article via Infotrieve]
  7. Dynlacht, B. D., Hoey, T., and Tjian, R. (1991) Cell 66, 563-576 [Medline] [Order article via Infotrieve]
  8. Zhou, Q., Lieberman, P. M., Boyer, T. G., and Berk, A. J. (1992) Genes & Dev. 6, 1964-1974 [Abstract]
  9. Poon, D., Bai, Y., Campbell, A. M., Bjorklund, S., Kim, Y. J., Zhou, S., Kornberg, R. D., and Weil, P. A. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8224-8228 [Abstract]
  10. Reese, J. C., Apone, L., Walker, S. S., Griffin, L. A., and Green, M. R. (1994) Nature 371, 523-527 [CrossRef][Medline] [Order article via Infotrieve]
  11. Meisterernst, M., Roy, A. L., Lieu, H. M., and Roeder, R. G. (1991) Cell 66, 981-993 [Medline] [Order article via Infotrieve]
  12. Lieberman, P. M., and Berk, A. J. (1994) Genes & Dev. 8, 995-1006 [Abstract]
  13. Kobayashi, N., Boyer, T. G., and Berk, A. J. (1995) Mol. Cell. Biol. 15, 6465-6473 [Abstract]
  14. Shykind, B. M., Kim, J., and Sharp, P. A. (1995) Genes & Dev. 9, 1354-1365 [Abstract]
  15. Koleske, A. J., and Young, R. A. (1995) Trends Biochem. Sci. 20, 113-116 [CrossRef][Medline] [Order article via Infotrieve]
  16. Chao, D. M., Gadbois, E. L., Murray, P. J., Anderson, S. F., Sonu, M. S., Parvin, J. D., and Young, R. A. (1996) Nature 380, 82-85 [CrossRef][Medline] [Order article via Infotrieve]
  17. Koleske, A. J., and Young, R. A. (1994) Nature 368, 466-469 [CrossRef][Medline] [Order article via Infotrieve]
  18. Kim, Y. J., Bjorklund, S., Li, Y., Sayre, M. H., and Kornberg, R. D. (1994) Cell 77, 599-608 [Medline] [Order article via Infotrieve]
  19. Ma, D., Watanabe, H., Mermelstein, F., Admon, A., Oguri, K., Sun, X., Wada, T., Imai, T., Shiroya, T., Reinberg, D., and Handa, H. (1993) Genes & Dev. 7, 2246-2257 [Abstract]
  20. Ozer, J., Moore, P. A., Bolden, A. H., Lee, A., Rosen, C. A., and Lieberman, P. M. (1994) Genes & Dev. 8, 2324-2335 [Abstract]
  21. Sun, X., Ma, D., Sheldon, M., Yeung, K., and Reinberg, D. (1994) Genes & Dev. 8, 2336-2348 [Abstract]
  22. DeJong, J., and Roeder, R. G. (1993) Genes & Dev. 7, 2220-2234 [Abstract]
  23. Yokomori, K., Zeidler, M. P., Chen, J. L., Verrijzer, C. P., Mlodzik, M., and Tjian, R. (1994) Genes & Dev. 8, 2313-2323 [Abstract]
  24. Yokomori, K., Admon, A., Goodrich, J. A., Chen, J. L., and Tjian, R. (1993) Genes & Dev. 7, 2235-2245 [Abstract]
  25. Cortes, P., Flores, O., and Reinberg, D. (1992) Mol. Cell. Biol. 12, 413-421 [Abstract]
  26. Ranish, J. A., Lane, W. S., and Hahn, S. (1992) Science 255, 1127-1129 [Medline] [Order article via Infotrieve]
  27. Ranish, J. A., and Hahn, S. (1991) J. Biol. Chem. 266, 19320-19327 [Abstract/Free Full Text]
  28. Kang, J. J., Auble, D. T., Ranish, J. A., and Hahn, S. (1995) Mol. Cell. Biol. 15, 1234-1243 [Abstract]
  29. Ozer, J., Bolden, A. H., and Lieberman, P. M. (1996) J. Biol. Chem. 271, 11182-11190 [Abstract/Free Full Text]
  30. Tan, S., Hunziker, Y., Sargent, D. F., and Richmond, T. J. (1996) Nature 381, 127-134 [CrossRef][Medline] [Order article via Infotrieve]
  31. Geiger, J. H., Hahn, S., Lee, S., and Sigler, P. B. (1996) Science 272, 830-836 [Abstract]
  32. Kim, Y., Geiger, J. H., Hahn, S., and Sigler, P. B. (1993) Nature 365, 512-520 [CrossRef][Medline] [Order article via Infotrieve]
  33. Kim, J. L., Nikolov, D. B., and Burley, S. K. (1993) Nature 365, 520-527 [CrossRef][Medline] [Order article via Infotrieve]
  34. Coulombe, B., Li, J., and Greenblatt, J. (1994) J. Biol. Chem. 269, 19962-19967 [Abstract/Free Full Text]
  35. Legrange, T., Kim, T. K., Orphanides, G., Ebright, Y., Ebright, R., and Reinberg, D. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 10620-10625 [Abstract/Free Full Text]
  36. Hori, R., Pyo, S., and Carey, M. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 6047-6051 [Abstract/Free Full Text]
  37. Li, R., and Botchan, M. R. (1993) Cell 73, 1207-1221 [Medline] [Order article via Infotrieve]
  38. Juo, Z. S., Chiu, T. K., Lieberman, P. M., Baikalov, I., Berk, A. J., and Dickerson, R. E. (1996) J. Mol. Biol. 261, 239-254 [CrossRef][Medline] [Order article via Infotrieve]
  39. Roberts, S. G., Ha, I., Maldonado, E., Reinberg, D., and Green, M. R. (1993) Nature 363, 741-744 [CrossRef][Medline] [Order article via Infotrieve]
  40. Nikolov, D. B., Chen, H., Halay, E. D., Usheva, A. A., Hisatake, K., Lee, D. K., Roeder, R. G., and Burley, S. K. (1995) Nature 377, 119-128 [CrossRef][Medline] [Order article via Infotrieve]
  41. Goodrich, J. A., Hoey, T., Thut, C. J., Admon, A., and Tjian, R. (1993) Cell 75, 519-530 [Medline] [Order article via Infotrieve]
  42. Nikolov, D. B., Hu, S. H., Lin, J., Gasch, A., Hoffmann, A., Horikoshi, M., Chua, N. H., Roeder, R. G., and Burley, S. K. (1992) Nature 360, 40-46 [CrossRef][Medline] [Order article via Infotrieve]
  43. Bryant, G. O., Martel, L. S., Burley, S. K., and Berk, A. J. (1996) Genes & Dev. 10, 2491-2504 [Abstract]
  44. Buratowski, S., and Zhou, H. (1992) Science 255, 1130-1132 [Medline] [Order article via Infotrieve]
  45. Lee, D. K., DeJong, J., Hashimoto, S., Horikoshi, M., and Roeder, R. G. (1992) Mol. Cell. Biol. 12, 5189-5196 [Abstract]
  46. Tang, H., Sun, X., Reinberg, D., and Ebright, R. H. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 1119-1124 [Abstract/Free Full Text]
  47. Greiner, D. P., Hughes, K. A., Gunasekera, A. H., and Meares, C. F. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 71-75 [Abstract/Free Full Text]
  48. Heyduk, E., and Heyduk, T. (1994) Biochemistry 33, 9643-9650 [Medline] [Order article via Infotrieve]
  49. Heyduk, T., Heyduk, E., Severinov, K., Tang, H., and Ebright, R. H. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 10162-10166 [Abstract/Free Full Text]

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.