Mot1 regulates the DNA binding activity of free TATA-binding protein in an ATP-dependent manner.

Mot1 is an essential Snf2/Swi2-related Saccharomyces cerevisiae protein that binds the TATA-binding protein (TBP) and removes TBP from DNA using ATP hydrolysis. Mot1 functions in vivo both as a repressor and as an activator of transcription. Mot1 catalysis of TBP.DNA disruption is consistent with its function as a repressor, but the Mot1 mechanism of activation is unknown. To better understand the physiologic role of Mot1 and its enzymatic mechanism, MOT1 mutants were generated and tested for activity in vitro and in vivo. The results demonstrate a close correlation between the TBP.DNA disruption activity of Mot1 and its essential in vivo function. Previous results demonstrated a large overlap in the gene sets controlled by Mot1 and NC2. Mot1 and NC2 can co-occupy TBP.DNA in vitro, and NC2 binding does not impair Mot1-catalyzed disruption of the complex. Residues on the DNA-binding surface of TBP are important for Mot1 binding and the Mot1.TBP binary complex binds very poorly to DNA and does not dissociate in the presence of ATP. However, the binary complex binds DNA well in the presence of the transition state analog ADP-AlF(4). A model for Mot1 action is proposed in which ATP hydrolysis causes the Mot1 N terminus to displace the TATA box, leading to ejection of Mot1 and TBP from DNA.

A critical step in the assembly of an active transcription complex at an RNA polymerase II promoter involves recruitment of TATA-binding protein (TBP) 1 and TBP-associated factors (1)(2)(3). TBP recruitment and activity are influenced by a large number of transcription factors and components of the general transcription machinery, many of which can interact directly with TBP (3)(4)(5)(6). MOT1 was uncovered in genetic screens for factors that repress transcription driven by a weak promoter (7)(8)(9)(10)(11). Consistent with its function as a repressor, Mot1 was isolated independently as an ATP-dependent factor that disrupts the TBP⅐DNA complex (12). Mot1 binds the TBP⅐DNA complex in vitro (12) and contacts both TBP and about 17 bp of DNA upstream of the TATA box (13). In the absence of DNA, Mot1 also dimerizes with TBP (13)(14)(15). In this report, we refer to the Mot1⅐TBP complex as the "binary" complex, and the Mot1⅐TBP⅐DNA complex is referred to as the "ternary" complex.
Mot1 homologs have been identified in many eukaryotes. The human homolog is BTAF1, which interacts with TBP (16,17) and catalyzes disruption of human TBP⅐DNA complexes (17). The insect homolog, the 89B helicase (18), may interact with TBP or TBP-related factor 1 (TRF1) in vitro (19). The Mot1 C terminus contains the conserved ATPase domain (7), whereas the Mot1 N terminus is responsible for TBP binding (19 -21). The structural basis for Mot1⅐TBP recognition is unknown, however, it was recently suggested that the Mot1 N terminus contains HEAT or ARM repeats, which compose a class of structurally related leucine-rich repeats (22)(23)(24). Structural studies have shown that HEAT and ARM repeats form two ␣ helices joined by a short loop (ARM repeats have a short additional ␣ helix), and these can stack upon each other to form a "superhelix" that provides an extensive surface for macromolecular interaction (22). Previous analysis of Mot1 deletion mutants indicated that an extended portion of the Mot1 N terminus is responsible for recognition of TBP (19,20). It has also been reported that, in solution, Mot1 is a non-globular monomer (15). Taken together, these data suggest a model in which Mot1 adopts an extended conformation that provides a large surface for interaction with TBP. To test the model, mutations were made in both Mot1 and TBP, and the effects on the Mot1⅐TBP interaction were determined. Because HEAT and ARM repeats are based mostly on hydrophobic interactions (22), it was expected that most polar residues in the N-terminal domain would not be essential, which we have found to be the case.
Mot1 is a member of the Snf2/Swi2 ATPase family (25)(26)(27). It has been suggested that at least some Snf2/Swi2 ATPases are processive molecular motors, acting by driving DNA translocation or rotation (28,29). The Mot1⅐TBP⅐DNA system has been used to test several theories about how these ATPases drive changes in protein⅐DNA interactions. Mot1 is not a helicase (13)(14)(15), nor does it travel long distances on DNA after TBP is removed from the TATA box (30). Catalysis of TBP⅐DNA disruption requires a grip by Mot1 on both upstream DNA and TBP, although the upstream DNA and the TBP⅐DNA complex can be conformationally uncoupled without impairing catalysis (13). These results indicate that Mot1 does not dissociate TBP⅐DNA by propagation of DNA twist or writhe through the TATA box. A similar result has been reported for the Snf2/Swi2 family member ISWI (31). It is possible that Mot1 interacts with the TATA box directly and in so doing alters its structure or that Mot1 uses ATP hydrolysis to disrupt TBP⅐DNA complexes via short-range tracking or ATP-driven insertion of Mot1 into the TBP⅐DNA interface. Alternatively, Mot1 may mediate TBP⅐DNA disruption by inducing a conformational change in TBP that deforms the DNA-binding surface of TBP.
Here we demonstrate that residues on the DNA-binding surface of TBP impair the interaction of TBP with Mot1, suggesting that Mot1 contacts the DNA-binding surface of TBP, and explaining why the Mot1⅐TBP binary complex binds DNA poorly compared with TBP alone. Binding of an ATP transition state analog locks the binary complex into a form in which the Mot1⅐TBP complex can bind DNA better than the nucleotidefree form of the Mot1⅐TBP complex. These results suggest that ATP hydrolysis causes a change in either the conformation of TBP or the interaction of Mot1 with the DNA-binding surface of TBP and that these ATP-driven conformational changes explain how Mot1 drives disruption of the TBP⅐DNA complex.

MATERIALS AND METHODS
mot1 Library Construction and Screening-Oligonucleotide primers flanking the EcoRI site (bp position 1026 in the MOT1 open reading frame (ORF)) and ClaI site (position 2092) were used to amplify ϳ1 kb of the MOT1 ORF using Taq polymerase under reduced fidelity conditions as described previously (32). The PCR-amplified DNA was digested with EcoRI and ClaI and cloned into an EcoRI-ClaI-gapped plasmid containing the rest of the MOT1 ORF under control of the GAL1 promoter on a CEN ARS plasmid bearing the LEU2 gene (20). Note that an additional ClaI site is present in the MOT1 ORF, but this second site is blocked from ClaI digestion by overlapping dam methylation. Six independent transformants were picked at random from the bacterial transformation of the primary ligation mix, and these were sequenced and found to contain ϳ1-bp change per kilobase (kb) of amplified DNA. Bacterial transformants containing the mutated DNA were then scraped en masse from agar plates, inoculated at high density into liquid media, and used in a large-scale plasmid purification prep. The resulting purified plasmids were then used to transform yeast strain AY29 (mot1⌬::TRP1, carrying plasmid pMR13 (MOT1ϩ URA3ϩ)) (20), which is otherwise congenic to YPH499 (33) by selection on synthetic complete media containing glucose but without leucine using standard techniques (34). Approximately 13,000 transformants were replica-plated to synthetic glucose-or galactose-containing media lacking leucine and incubated at 30°C for 3-5 days. Comparison of the glucose and galactose-containing plates did not reveal any GAL1-inducible alleles of MOT1, which caused slow growth in the presence of wild-type MOT1. Colonies were then replica-plated from galactosecontaining media to media containing galactose and 5-fluoroorotic acid (34) to select for loss of the URA3-marked plasmid containing the wild-type MOT1 gene. Approximately half of the transformants did not survive the 5-fluoroorotic acid selection, indicating that these strains harbored alleles of MOT1, which do not support growth in the absence of wild-type MOT1. The remaining viable strains were screened for temperature-sensitive growth defects by replica plating to synthetic galactose plates minus leucine and incubation at 30°C and 35°C. Temperature-sensitive strains were re-streaked, and the plasmids were isolated and re-transformed to the MOT1 deletion strain to confirm the plasmid-linked temperature-sensitive (ts) phenotype. Candidate genes were then sequenced through the entire EcoRI-ClaI region of the ORF, and the mutant fragments were subcloned to a new plasmid backbone containing the remainder of the MOT1 gene to be sure that mutations in the EcoRI-ClaI DNA fragment were responsible for the phenotypes observed.
Site-directed MOT1 Mutants-Site-directed mutagenesis was performed using synthetic oligonucleotides and either overlapping PCR or the Stratagene QuikChange kit, according to the instructions provided by the manufacturer. Each mutation was engineered to encode a change in a restriction site (either introduction of a new site or loss of an existing site) to facilitate subcloning. Candidate transformants containing the correct restriction sites were then sequenced completely in a region that overlaps a DNA fragment with convenient restriction sites. The sequenced DNA fragment was then sub-cloned to LEU2 CEN ARS plasmids derived from pRS315 (33) that contain the MOT1 ORF driven by the GAL1 promoter or by a 448-bp fragment of the MOT1 promoter. All constructs encode a Mot1 derivative with the Py tag (35) appended to the N terminus to facilitate quantitation by Western blotting and purification using antibody-coupled beads (20,35). Additional details regarding plasmid construction are available upon request. Plasmids containing the site-directed alleles were transformed into AY29 yeast cells (see above), and the ability of the constructs to support viability was assessed by plasmid shuffling using standard techniques (34). Strains harboring alleles under control of the MOT1 promoter were analyzed for growth defects on synthetic media without leucine and containing raffinose, galactose, or glucose as the carbon source. Strains harboring alleles under control of the GAL1 promoter were streaked to galactose-containing plates to induce expression prior to plasmid shuffling. Growth of strains was compared with congenic wild-type cells by incubation at 16°C, 30°C, 32°C, and 35°C.
Purification of Recombinant TBP and TBP Mutants-Recombinant full-length TBP and TBP mutants expressed under the control of the T7 promoter as a fusion with N-terminal six-histidine tag were obtained by transformation of BL21(DE3) Escherichia coli cells with the appropriate plasmid expression vectors (13,36). Cells were inoculated into 1 liter of yeast extract Tryptone (YT) media containing 100 g/ml ampicillin or 30 g/ml kanamycin at 37°C and were grown to an optical density at 600 nm of 0.7-1.0. Isopropyl-␤-D-thiogalactopyranoside was added (0.5 mM final concentration), and the cells were incubated at 37°C for 3 h to allow protein expression. Cells were harvested and resuspended in buffer I (40 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 2 M benzamidine, 2 M pepstatin, 0.6 M leupeptin, and 2 g/ml chymostatin) containing 5 mM imidazole and lysed by sonication. After sonication, the cleared lysate was incubated at 4°C for 1 h with 0.2 ml of nickelnitrilotriacetic acid-agarose (Qiagen) pre-equilibrated with buffer I, which included 5 mM imidazole. The mixture was then loaded into a column, and the resin was washed with 10 ml buffer I plus 5 mM imidazole and subsequently with 5 ml of buffer I containing 20 mM imidazole. Finally, the bound protein was eluted with buffer I containing 200 mM imidazole. The yield was quantitated by Bradford assay (Bio-Rad), and the purity was assessed by Coomassie Blue staining of 10% protein gels. Based on the Coomassie Blue staining, the proteins were estimated to be about 90% pure.
Native Gel Electrophoresis-For detection of TBP by native gel electrophoresis (Figs. 6C, 6D, and 7B), full-length TBP and Mot1 were incubated in binding buffer (13) containing 120 mM KCl and 12 mM HEPES buffer, pH 7.6 (37), using proteins at the concentrations indicated in figure legends. The gels in Figs. 6C and 7B were run with the electrodes reversed: samples were loaded on the side of the positive electrode, and run toward the negative electrode. The gel was, however, pre-run for Ͼ50 min with the electrodes connected in the usual fashion before loading. Following electrophoresis, the gels were boiled in 1% SDS for 1 min, then transferred to Immobilon and TBP was detected using TBP antiserum. The TBE gel shift assay in Fig. 8A was performed as described previously (38). Gel shift assays were otherwise performed as described previously (13) using 5 nM core domain TBP (gift of J. Geiger) or full-length TBP with minor modifications as indicated. Synthesis and labeling of the 36-and 17-bp DNAs, and preparation of the radiolabeled 100-bp adenovirus major late promoter fragment, was as previously described (13). DNA concentration was about 0.5 nM in the reactions. The concentration of Mot1 needed to bind 50% of the TBP⅐DNA complex is ϳ5 nM (13). The concentration of Mot1 used was estimated from this activity and is indicated in the figure legends. ATP was used at between 5 and 100 M. ADP was used at 100 M. NaF was used at 2.5 mM. AlCl 3 was used at 10 M (39). Bur6 was used at 13 nM and Ydr1/Ncb2 at 60 nM (38); both proteins were a gift of G. Prelich.
Purification of Mot1 and Pull-down Assays-Mot1 was expressed and purified from yeast using antibody-coupled beads exactly as described previously (20). The antibody-coupled beads were prepared using Py monoclonal antibody that recognizes the Mot1 epitope tag (20), which was prepared at the University of Virginia Lymphocyte Culture Center. For detection of TBP binding to immobilized Mot1, Mot1-coupled beads were equilibrated with buffer T-60 (30 mM Tris (pH 8.0), 5 mM magnesium chloride, 0.1% Brij-58, 1 mM dithiothreitol, protease inhibitors plus 60 mM potassium chloride). One hundred nanograms of full-length recombinant yeast TBP (or TBP mutant) was added in 500 l of buffer T-60, and the reaction was incubated for 30 min at room temperature. After binding, the unbound material was collected and the beads were washed with buffer T containing increasing concentrations of KCl; samples marked "Eluate" were collected in T-1000. The eluted proteins were precipitated with acetone, and TBP present in the eluates was detected by Western blotting using rabbit polyclonal anti-TBP antisera.
Preparation of GST-TBP and Mot1 Binding to GST-TBP-One-liter cultures of DH10B bacterial cells containing plasmid pGEX-1 (Amersham Biosciences) or a plasmid expressing GST fused to full-length yeast TBP (kindly provided by Ron Reeder) were grown in YT medium at 37°C to an optical density at 600 nm of 0.7-1.0. Isopropyl-␤-D-thiogalactopyranoside was added (1.0 mM, final concentration), and the cells were incubated at 37°C for an additional 3 h. Cells were harvested by centrifugation and resuspended in 20 ml of buffer T (30 mM Tris-HCl, pH 8.0, 2 M pepstatin A, 1 mM phenylmethylsulfonyl fluoride) containing 150 mM KCl (T-150 buffer). Cells were lysed by sonication, and debris was removed by centrifugation. After centrifugation, 0.5 ml of GST lysate or 1.5 ml of GST-TBP lysate was incubated at 4°C for 1 h with 20 l of glutathione-agarose equilibrated with buffer T containing 60 mM KCl (T-60). The agarose was washed three times with 1 ml of T-150 buffer, and the entire 20-l sample of agarose-bound material was used for testing the binding of Mot1. 10 ng of the eluted Mot1 protein obtained from yeast overexpression strains (see above) was added to a 20-l suspension of GST or GST-TBP agarose in T-60 buffer and incubated on a roller for 1 h at 4°C. The agarose was washed once with 0.6 ml of T-60 buffer, then an elution step was carried out with 0.6 ml of T-60 buffer with 5 mM MgCl 2 , with or without 50 M ATP and with or without 1 nM TATA sequence DNA. Eluted proteins were precipitated with acetone for analysis by Western blotting using the Py antibody (35), which recognizes the N-terminal epitope tag.
Sequence Analysis-Blocks of conserved sequences in the Mot1 N terminus were identified with a set of Mot1 homologs found in Entrez protein sequence data bank (available at www.ncbi.nlm.nih.gov/entrez/ query.fcgi?dbϭProtein) (from Homo sapiens (accession number AAC04573), Arabidopsis thaliana (T47857), Saccharomyces cerevisiae (P32333), Schizosaccharomyces pombe (T40642), and Drosophila melanogaster (AAF55260)) using the MACAW (40) and ClustalW algorithms (available at www.ibc.wustl.edu/service/msa/index.html). HEAT repeats in Mot1 reported by previous authors (23,24) included five in the B block. We observed sequence similarity of the second B block HEAT repeat to sequences immediately downstream using MACAW. This had not been found in the original sequence analysis, and Fig. 1B therefore includes an additional HEAT repeat in the B block, for a total of six.

RESULTS
Four Conserved Regions in Mot1 N Terminus-Alignment of Mot1 homologs revealed conserved blocks outside of the ATPase, which we designate A-D (Fig. 1A). These blocks were not found in any protein except Mot1 or its homologs. The A and B blocks in the human and yeast proteins are about 40% identical. For example, Fig. 1B shows the sequence of the S. cerevisiae Mot1 A block; the asterisks indicate residues that are identical in the yeast and human proteins. Mot1 contains a series of HEAT repeat sequences dispersed throughout the N terminus (23,24). Remarkably, the four conserved N-terminal domains of Mot1 coincide with the positions of the HEAT repeats ( Fig. 1, A and B; the brackets in Fig. 1B indicate where two HEAT repeats fall within the A block).
Temperature-sensitive Alleles of mot1-PCR-based mutagenesis was used to introduce random changes in the MOT1 open reading frame in the region between codons for Arg-345 and Asn-697. This region was chosen because previous deletion analysis indicated an important role for residues in this region in TBP recognition (19,20). A CEN ARS plasmid library expressing mutagenized mot1 under GAL1 control was used, because genes that express catalytically defective mot1 were expected to be dominant inhibitors of cell growth (20). Library construction and screening are described under "Materials and Methods." Note that 35°C was chosen for the non-permissive temperature, because the wild-type MOT1ϩ strain used in these studies is itself somewhat growth-impaired at temperatures above 35°C (not shown). The alleles isolated are recessive, and most of these contain multiple base pair changes (Table I). A single amino acid change, L383P, is responsible for the temperature-sensitive (ts) growth phenotype of a strain harboring mot1-41, because the same mutation in mot1-42 conferred the same phenotype. However, more than one amino acid change is required for the ts phenotypes conferred by mot1-71 and mot1-81, because no single amino acid change encoded by these alleles resulted in the conditional phenotype; several pairwise combinations of mutations in conserved residues also failed to confer a ts phenotype (not shown). The mot1-14 phenotype likely results from a low level of translational by-pass of the premature stop codon substituted for Trp-496, because deletion of the mot1 open reading frame downstream of this stop codon is lethal (not shown).
Growth phenotypes of strains carrying the GAL1-driven alleles are summarized in Table I. Comparison of strain growth by serial dilution spot assay demonstrated that, compared with wild-type cells, the mot1 strains displayed growth defects of ϳ100 -1000-fold when incubated at 35°C (not shown). Growth phenotypes of these mot1 strains were similar regardless of whether the alleles were expressed under control of the GAL1 or MOT1 promoters. Western blot analysis of whole cell extracts from cells grown at 30°C, using an antibody that recognizes epitope-tagged versions of these proteins, demonstrated that proteins encoded by mot1-41, mot1-71, and mot1-81 were expressed at wild-type levels, whereas full-length protein encoded by mot1-14 was nearly undetectable ( Fig. 2A). Mot1 protein level in the mot1-42 strain is intermediate ( Fig. 2A, Wild-type and mutant Mot1 proteins were purified from yeast overexpression strains using antibody-coupled beads (20). The purified proteins were then tested in gel mobility shift assays for the ability to bind TBP⅐DNA complexes and for ATP-dependent TBP⅐DNA disruption activity (20). As shown in Fig. 2B (lanes 1-7), addition of wild-type Mot1 led to formation of Mot1⅐TBP⅐DNA ternary complexes that were disrupted in the presence of ATP. The Mot1-41, Mot1-42, Mot1-71, and Mot1-81 proteins did not stably bind to TBP⅐DNA complexes. These proteins also failed to disrupt TBP⅐DNA complexes in the presence of ATP even when severalfold more protein was used than was required for wild-type Mot1 to quantitatively supershift and disrupt the TBP⅐DNA complexes formed under these conditions (Fig. 2B, lanes 8 -13, and Fig. 2C). Thus, the defects in cell growth resulting from Mot1-41, Mot1-42, Mot1-71, or Mot1-81 can be explained by general defects of these proteins in TBP⅐DNA recognition.
Alanine Scanning Mutations in the MOT1 A Block-To map the surface of Mot1 required for TBP binding, extensive mutagenesis of the N-terminal conserved regions was undertaken. While this work was in progress, it was reported that the A and B blocks contain HEAT (or ARM) repeats (22)(23)(24). These repeats stack via hydrophobic interactions to form an extended, helical structure (22); thus, alanine scanning of polar residues should not affect the overall fold but could inhibit polar interactions. Mot1 missing the entire A block (Mot1-260; deletion of amino acids 1-98) does not support cell viability (Table II) even though this N-terminally truncated protein is expressed at wild-type levels (not shown). The A block is thus essential for Mot1 function in vivo. The Mot1-260 protein is also defective for formation of Mot1⅐TBP⅐DNA ternary complexes and ATPdependent disruption of TBP⅐DNA in vitro (Fig. 3A, lanes 3 and  4). Similarly, deletion of the "linker" connecting A and B blocks generated a non-functional protein in vivo and in vitro (mot1-274; Fig. 3A and Table II) suggesting that important residues are located within the linker or that the A and B blocks must be appropriately positioned for Mot1 to function.
Site-directed A block mutant alleles were constructed on low copy plasmids under control of the MOT1 or GAL1 promoter and introduced into a yeast strain containing a deletion of the chromosomal copy of MOT1. All of the conserved charged and polar residues within the A block were mutated to alanine either singly or in clusters, and, remarkably, none of these residues were found to be essential for Mot1 function in vivo, even though many of these residues are conserved across species (see Fig. 1B, data are summarized in Table II). Strains carrying each of the MOT1-driven A block alleles were also screened for growth defects at 16°C, 32°C, and 35°C but no ts or cold-sensitive phenotypes were observed (not shown). These strains also display no growth defects on synthetic or rich media with glucose, galactose, or raffinose as the carbon source (not shown).
In contrast to the normal growth observed with the A block alanine mutations when expressed under control of the MOT1 promoter, several A block mutant alleles displayed a severe dominant-negative phenotype when expressed from the GAL1 promoter. GAL1-driven MOT1 is expressed at a 20 -50-fold higher level than MOT1 under control of the MOT1 promoter (Fig. 3B, lane 1 versus 2). As summarized in Table II, GAL1expressed alleles of MOT1-encoding mutations at the extreme N terminus severely inhibited cell growth when cells were grown on galactose in the presence of wild-type Mot1. The most severe defect was seen in MOT1-101 cells (Table II). MOT1-101 and wild-type MOT1 were expressed at equivalent levels under GAL1 control (Fig. 4B), and severe growth defects were observed on plates containing galactose or galactose plus raffinose (not shown), indicating that these alleles do not simply confer an inability of cells to metabolize galactose. In the absence of the wild-type MOT1 gene, cells expressing MOT1-101 were inviable on galactose-containing media, cells expressing mot1-204, mot1-205, or mot1-206 grew more slowly than wildtype cells at 30°C, and mot1-204 and mot1-205 conferred ts growth at 35°C (not shown). The lethality induced by GAL1driven MOT1-101 is due to elevated expression levels of this protein, because cells grew well with MOT1-101 as the sole source of Mot1 when the allele was expressed under control of the normal MOT1 promoter (Table II).
Alleles of MOT1 that encode proteins that recognize TBP but are defective in ATP-dependent TBP⅐DNA disruption exert dominant-negative effects on cell growth (41). This is due to interference with TBP function, because these dominant-negative phenotypes can be suppressed by overexpression of SPT15, which encodes TBP (41). To determine if the dominantnegative A block mutants interfere with TBP function in vivo, high copy plasmids expressing SPT15 were introduced into strains expressing the dominant-negative A block allele MOT1-101. As shown in Fig. 3C, the lethality induced by GAL1-driven MOT1-101 can be suppressed by SPT15 overexpression. Side-by-side comparisons (not shown) demonstrate that overexpression of SPT15 does not fully restore growth of these strains to wild-type rates, but these results suggest that the lethality induced by these A block mutations can be explained, at least in part, by interference with normal TBP function in vivo. SPT15 overexpression was unable to suppress the growth defect in the GAL1-MOT1-101 cells in which MOT1-101 was the only source of Mot1 (not shown). This suggests that elevated levels of Spt15 suppress MOT1-101 by interacting with the encoded mutant protein and thereby allowing wild-type Mot1 to function. As shown in Fig. 3D, Mot1-101 does recognize and dissociate TBP⅐DNA complexes in vitro, but the affinity of Mot1-101 for TBP⅐DNA complexes is reduced  2. Analysis of temperature-sensitive mot1 alleles. A, Western blot (Py monoclonal antibody) analysis of Mot1 protein levels present in whole cell extracts prepared from strains carrying the indicated mot1 allele. "wt" refers to wild-type Mot1. In extract from the strain labeled "vector," Mot1 is present but is untagged. In lanes 1-6, the indicated mot1 alleles were expressed from the GAL1 promoter.  Table II. Mutation of Glu-308, Arg-310, and His-311 resulted in a recessive loss-of-function allele (mot1-102; Table  II). mot1-102 is expressed at wild-type levels (Fig. 4B), and this protein does not detectably recognize or disrupt TBP⅐DNA complexes (Fig. 4A). Mutation of conserved residues Asp-361 or Asp-365 also results in complete loss of Mot1 function in vivo (Table II). The Mot1-104 protein, which contains both of these amino acid changes is expressed at wild-type levels (Fig. 4B), and, like Mot1-102, does not detectably bind or disrupt TBP⅐DNA complexes in vitro (Fig. 4A). Two alleles with wildtype in vivo function that encode changes in highly conserved residues were analyzed biochemically and found to recognize TBP⅐DNA complexes and support ATP-dependent TBP⅐DNA disruption equivalently to wild-type Mot1 (Mot1-103 in Fig. 4A, Mot1-216 in 4C). Thus, the ability of the MOT1 alleles to support growth and the abilities of the encoded proteins to support TBP⅐DNA disruption are correlated.
TBP Residues That Participate in Mot1 Binary and Ternary Complex Formation-The Mot1⅐TBP binary complex binds DNA poorly (13). This suggests either that Mot1 interacts with the DNA-binding surface of TBP or that Mot1 induces a conformational change in TBP that affects the ability of TBP to bind to DNA. These results also suggest that Mot1 contacts TBP differently depending on whether TBP is bound to DNA. The N terminus of a human Mot1 homolog, BTAF1, also binds to TBP and can inhibit TBP binding to DNA (21). To better define how Mot1 recognizes TBP, TBP mutants were tested in vitro for the ability to interact with Mot1 in the absence of DNA. TBP mutants that retain DNA binding activity were also tested to determine if Mot1 could catalyze disruption of their interaction with DNA. Mot1 was loaded onto antibody-coupled beads as previously described (20), the beads were incubated with full-length TBP and washed, and the TBP association with the beads was assayed by Western blotting using TBP antibodies. TBP was retained on beads loaded with Mot1, whereas TBP binding to beads alone was nearly undetectable (Fig. 5A, lanes 5 and 9 versus lane 3). Control experiments established that the Mot1⅐TBP binary interaction was insensitive to ethidium bromide and DNase I, indicating that the association was not mediated by contaminating DNA (Fig. 5A,  lanes 5 versus 7 and 9 versus 11). Because Mot1 used in these experiments was obtained from a yeast overexpression system, we also established that there was no contaminating TBP in the affinity-purified Mot1 preparation, and the TBP retained by the Mot1 beads therefore resulted from the interaction of Mot1 with the recombinant TBP added to the reactions (Fig.  5A, lanes 12-15).
Two mutants with solvent-exposed amino acid changes in the same ␣ helix (helix 2) on the "top" convex surface of TBP were tested for interaction with Mot1 in this assay. As shown in Fig. 5B, TBP K138L and K145L were both defective for interaction with Mot1. TBP Lys-145 was previously shown to be critical for Mot1 recognition of TBP⅐DNA complexes (38), and TBP K133L,K138L was shown to be defective for Mot1-catalyzed disruption (12). TBP K138L⅐DNA complexes are not sta- bly bound Mot1 (Fig. 5D), demonstrating that mutation of either Lys-138 or Lys-145 alone is sufficient to block recognition by Mot1. Thus, these residues on the convex surface of TBP are required for interaction with Mot1 in both the presence and absence of DNA. TBP Lys-127 is located at the extreme N terminus of helix 2 near the upstream edge of the TBP DNAbinding surface (see Fig. 9A). Whereas TBP K127L is defective for interaction with Mot1 in the absence of DNA (Fig. 5B), Mot1 can stabilize the interaction of TBP K127L with DNA to some extent, and the Mot1⅐TBP K127L⅐DNA ternary complex dissociates in the presence of ATP (Fig. 5E). This residue may define a difference in the architecture of the Mot1⅐TBP and Mot1⅐TBP⅐DNA complexes, or alternatively, this ternary complex may fall apart during ATP hydrolysis simply because TBP K127L binds DNA poorly. Three TBPs with mutations on the DNA-binding surface were also tested for interaction with Mot1. As shown in Fig. 5B, TBP N159D retained the ability to interact with Mot1, whereas TBP V71E and TBP V161E do not interact detectably with Mot1 in this assay. The Mot1 N terminus binds weakly to TBP⅐DNA (20), and the Mot1 N terminus is also sufficient for formation of the Mot1⅐TBP binary complex (Fig. 5C). As was observed with the full-length Mot1 protein, there was no detectable binding of the Mot1 N terminus to TBPs with mutations in critical residues on the convex (K145L) or concave, DNA binding (V71E) surface of TBP. These results are consistent with and extend previously published results (21) and suggest that the inability of Mot1⅐TBP complexes to bind to DNA (13) is due to a direct interaction between Mot1 and the TBP binding surface. Remarkably, this would imply that the Mot1 N terminus embraces TBP via an extensive surface, making specific contacts with TBP simultaneously on opposite sides of the molecule.
Conformational Change of Mot1⅐TBP Binary Complex Induced by ATP-The Mot1⅐TBP binary complex does not detectably bind DNA in vitro, but it does hydrolyze ATP (20). Addition of ATP to pre-formed Mot1⅐TBP complexes allows TBP⅐DNA complexes to assemble on a DNA template that is too short to support Mot1-catalyzed disruption (13). One interpretation of these results is that Mot1 dissociates from TBP in the presence of ATP. Other experimental approaches have led to the conclusion that ATP does not induce the Mot1⅐TBP complex to dissociate (15), suggesting that disruption of Mot1⅐TBP binary complexes requires both ATP and DNA. To test this idea, Cultures of cells were grown to mid-log in raffinose-containing medium then induced with the addition of galactose to 2% for 2 h prior to harvest. "Vector" refers to extract from cells harboring plasmid with no epitope-tagged MOT1 gene. C, gel mobility shift analysis as in A using purified Mot1-216, which displays ternary complex formation and TBP⅐DNA disruption activity equivalent to wild-type Mot1. three different experimental approaches were compared directly. In the first experiment, ATP and a 17-bp TATA DNA sequence (too short to support Mot1 binding, see "Materials and Methods" and Ref. 13) were added to a reaction containing the pre-formed Mot1⅐TBP binary complex. As shown in Fig. 6A  (lanes 7 and 8), ATP induced the dissociation of Mot1⅐TBP binary complexes and formation of TBP⅐DNA complexes when a short DNA template was added to the reaction. Using a standard 36-bp DNA template that does support Mot1 action (Fig. 6A, lanes 1-6), Mot1 can load onto pre-formed TBP⅐DNA complexes and disrupt them using ATP, but, as expected, no TBP⅐DNA complexes were detected when the DNA and ATP were added to Mot1⅐TBP complexes, because any TBP loaded onto this DNA template was dissociated by Mot1. Consistent with these and previously published results (15), Mot1 bound to GST-TBP beads was not released in the presence of ATP alone (Fig. 6B). In contrast, however, ATP and DNA catalyzed release of less than half the Mot1 from GST-TBP beads (Fig.  6B, lane 5). The simplest interpretation of these results is that ATP and DNA can induce dissociation of the Mot1⅐TBP binary complex, but that tethering TBP to agarose beads impairs the catalytic activity of Mot1. Similar results were obtained in a reciprocal experiment using Mot1 bound to agarose beads and TBP in solution (not shown).
To better define the effect of ATP on the Mot1⅐TBP binary complex, a non-denaturing gel electrophoresis assay was used, but TBP and Mot1 were monitored by Western blotting rather than using radiolabeled DNA as in a conventional gel shift experiment. Under these conditions, free TBP was positively charged and entered a gel run toward the negative electrode (37). As shown in Fig. 6C (lanes 1-6), addition of Mot1 diminished the amount of free TBP that entered the gel. Mot1 was also incubated with TBP K138L, a TBP mutant that is not recognized by Mot1 (see Fig. 5). As shown in Fig. 6C (lanes  9 -14), the amount of free TBP K138L was not diminished by addition of Mot1 so the decrease in the amount of TBP detected when Mot1 was added is not trivially due to degradation of TBP. Although TBP is slightly positively charged, Mot1 is predicted to have a slight negative charge under these conditions and the bulky Mot1⅐TBP binary complex is apparently nearly uncharged and did not enter gels run toward either the positive or the negative electrode. The negatively charged FIG. 5. TBP DNA-binding surface is critical for Mot1 binding to TBP. A, TBP interaction with Mot1 bound to agarose beads. Agarose beads with or without Mot1 (as indicated) were incubated with recombinant full-length TBP. The unbound flow-through (FT) and bead-bound materials (Eluate) were analyzed by Western blotting using a rabbit polyclonal antibody directed against TBP. DNase I or ethidium bromide (EtBr) were included in the wash buffers as indicated. In the reaction analyzed in lanes 14 and 15, no recombinant TBP was added; the absence of detectable TBP signal indicates that TBP did not contaminate the Mot1 preparation used for these studies. B, interaction of recombinant TBP or TBP mutants with agarose beads or Mot1 beads. The analysis was performed as in A using wild-type TBP or mutants as indicated. Note that wild-type TBP and TBP N159D are the only proteins that bound detectably to Mot1 beads (lane 5). C, binding of wild-type TBP or TBP mutants to beads alone (vector) or Mot1 N-terminal fragments bound to beads. Mot1-1280 is a Mot1 fragment with residues 1-1280, and Mot1-800 has the first 800 residues of Mot1 (see Fig. 1A). Analysis was performed as in A and B of this figure. D, gel mobility shift analysis was performed using radiolabeled DNA, Mot1, and the indicated TBPs as in Fig. 2C. Lane 1 shows position of free DNA. Lanes 2-8 each contain 2.5 nM purified recombinant full-length wild-type TBP. Lanes 9 -15 each contain 2.5 nM purified recombinant TBP K138L. Relative amounts of purified Mot1 were added as indicated, where 1 unit (lane 3) is ϳ5 nM. Note that TBP K138L⅐DNA complexes are unaffected by Mot1. E, gel mobility shift analysis as in D but using TBP K127L where indicated. Lane 1 shows position of free DNA. Reactions in lanes 2-8 each contained 2.5 nM purified recombinant full-length wild-type TBP, and reactions in lanes 9 -15 each contained 25 nM purified recombinant TBP K127L (a longer exposure is shown than in panel D). Relative amounts of purified Mot1 were added as in D. TBP K127L is defective for DNA binding, and the TBP K127L⅐DNA complex was barely detectable under these conditions. However, Mot1 stabilized TBP K127L binding to DNA (note ternary complex in lanes 10 -12), and the complex dissociated in the presence of ATP (lanes 13-15).
TBP⅐DNA complex could be detected, however (Fig. 6D, lane 2). As expected, the Mot1⅐TBP⅐DNA ternary complex was not formed on the 17-bp DNA used (Fig. 6D, lane 3). Importantly, addition of ATP to Mot1⅐TBP binary complexes did not result in release of free TBP (Fig. 6C, lane 6 versus 5). However, addition of ATP and DNA to pre-formed Mot1⅐TBP complexes resulted in the appearance of the TBP⅐DNA complex (Fig. 6D, lane 6), consistent with the results in Fig. 6A. We conclude that while the Mot1⅐TBP binary complex hydrolyzes ATP, addition of ATP alone does not induce the complex to fall apart. However, the Mot1⅐TBP binary complex can be dissociated in a reaction that contains both ATP and DNA. Mot1⅐TBP complexes bound to agarose beads do not support Mot1 catalytic activity.
A Transition State ATP Analog Facilitates Loading of Mot1⅐TBP Complexes onto DNA-Because ATP does not induce Mot1⅐TBP binary complex dissociation but does facilitate TBP binding to DNA, we considered the possibility that locking the Mot1 ATPase into a conformational state somewhere along the catalytic path could generate a Mot1⅐TBP binary complex with enhanced DNA binding activity. This was tested using ADP aluminum fluoride (ADP-AlF 4 ), which binds to ATP-binding sites and mimics the presumed transition state of ATP during hydrolysis (39). ADP-AlF 4 does not cause disruption of the Mot1⅐TBP⅐DNA ternary complex (Fig. 7A, lane 5 versus 3), although ADP-AlF 4 appears to bind to the Mot1 ATP-binding site, because preincubation of Mot1 with ADP-AlF 4 prevents Mot1 from utilizing ATP added subsequently (Fig. 7A, lanes  11-14). Disruption of TBP⅐DNA complexes by Mot1 therefore requires ATP hydrolysis or perhaps multiple rounds of ATP hydrolysis. Consistent with the results in Fig. 6A, little Mot1⅐TBP⅐DNA ternary complex was detected when Mot1 and TBP were preincubated prior to addition of DNA (Fig. 7A, lanes  6 versus 3). However, addition of ADP-AlF 4 to pre-formed Mot1⅐TBP binary complexes allowed the binary complexes to load onto DNA (Fig. 7A, lanes 8 versus 3 and 6). Interestingly, addition of ADP-AlF 4 does not cause the Mot1⅐TBP binary complex to dissociate (Fig. 7B), suggesting that the Mot1⅐TBP⅐ADP-AlF 4 complex does not require interactions between Mot1 and the DNA-binding surface of TBP for stability or that Mot1 modulates the DNA binding activity of TBP exclusively by directing conformational changes in TBP. Thus, ADP-AlF 4 can convert the conformation of the Mot1⅐TBP binary complex into a form capable of binding DNA. Because ATP hydrolysis does not cause the binary complex to dissociate, these results support the hypothesis that one stage of the Mot1 ATP hydrolysis cycle opens or activates the binary complex to DNA binding.
Interaction of Mot1 with NC2⅐TBP⅐DNA-Because mutations on the DNA-binding surface of TBP impair interaction with Mot1, and ATP or ADP-AlF 4 can cause the TBP DNA-binding surface in the binary complex to become accessible to DNA, one simple model is that a portion of the Mot1 N terminus contacts the TBP DNA-binding surface and this interaction is transiently disrupted during the ATP hydrolysis cycle. NC2 specifically recognizes the "underside" of the TBP⅐DNA complex (42), so NC2 might block loading of or catalysis by Mot1 by prevent-  2-7) were incubated in the absence (lanes [1][2][3][4][5][6] or presence (lane 7) of ATP. An identical series of reactions were run in parallel using TBP K138L rather than wild-type TBP (lanes 9 -14). The reactions were loaded at the top onto non-denaturing polyacrylamide gels and electrophoresed with electrodes connected as shown. Following electrophoresis, Western analysis was performed to detect TBP or TBP K138L. The band represents monomeric TBP (37) ing Mot1 from interacting with the concave surface of TBP. There is also a striking overlap of the gene sets whose transcription is controlled by NC2 and Mot1, suggesting a mechanistic interplay between them at specific promoters (11,(43)(44)(45). To address the biochemical interplay of Mot1 and NC2 in vitro, gel mobility shift experiments were performed. The gel mobility shifts of the Mot1⅐TBP⅐DNA and NC2⅐TBP⅐DNA complexes were readily distinguished with the NC2⅐TBP⅐DNA complex migrating just slightly more slowly than the TBP⅐DNA complex (Fig. 8A). Addition of both Mot1 and NC2 resulted in the appearance of a new species (Fig. 8A, lane 6). The same results were observed in both TBE gels (Fig. 8A), which favor the NC2⅐TBP⅐DNA shift, and in TG gels (Fig. 8B, compare band marked by the bracket with that marked by the asterisk), which stabilize the TBP⅐DNA shift but destabilize the NC2⅐TBP⅐DNA shift. These results show that Mot1 and NC2 do not compete for TBP binding; rather, they cooperate to form Mot1⅐NC2⅐ TBP⅐DNA quaternary complexes. In addition, ATP caused disruption of the quaternary complex (Fig. 8A, lane 6 versus 7; Fig.  8B, lane 6 versus 7), indicating that NC2 also provides no barrier to Mot1-catalyzed TBP⅐DNA disruption.

DISCUSSION
Mot1 is an essential, conserved yeast protein (7) that interacts genetically and biochemically with TBP (14,41). Mot1 catalysis of TBP⅐DNA disruption can explain its role as a repressor of transcription (43), but how Mot1 activates transcription and the mechanism of the TBP⅐DNA disruption reaction are unknown. The results in this paper provide mechanistic insight into how the Mot1 ATPase is used to drive TBP⅐DNA disruption in vitro, and in vivo analysis of Mot1 mutants demonstrates a close correlation between the ability of Mot1 to catalyze TBP⅐DNA disruption and the ability to provide the essential function of Mot1 in vivo. These results also explain previous data demonstrating inhibition of TBP DNA binding by Mot1. Furthermore, the role of ATP in opening the Mot1⅐TBP binary complex to DNA binding suggests a model, discussed below, for how Mot1 catalyzes TBP⅐DNA disruption.
Leucine Repeats in the Mot1 N Terminus-The leucine repeats of the Mot1 N-terminal domain have been identified as either HEAT (24) or ARM (22) repeats. In either case, the leucine repeats of Mot1 coincide with the blocks of conserved sequence in the N terminus (Fig. 1). There is no structural information about them, but these results suggest that the Mot1 N terminus probably adopts an extended conformation, similar to importin ␤ (46,47) or ␤ catenin (48). The Mot1 N terminus is sufficient for TBP binding (Fig. 5C) and is necessary for activation of the ATPase (20). Karyopherin 114, an importin ␤ family member, binds TBP (49,50), so there may be other TBP-hydrophobic repeat interactions.
A direct test of the ARM/HEAT repeat model for Mot1 is not possible because of the limited structural information on Mot1 and the large size of the Mot1 N terminus. Hydrophobic residues are also predicted to play important roles in both stabilization of interactions between leucine repeats and interaction with TBP, but Mot1 proteins with mutations in hydrophobic residues would also be expected to be defective, because mutation of residues in the hydrophobic core of the protein could  1, 4, and 7) or presence of ATP (lanes 3 and 6) or ADP-AlF 4 (lanes 2 and 5). Proteins were incubated with ATP or ADP-AlF 4 for a total of 30 min. In reactions with both Mot1 and TBP, Mot1 was incubated with ATP or ADP-AlF 4 for 20 min, followed by addition of TBP for 10 min. The reactions were loaded onto non-denaturing polyacrylamide gels and electrophoresed as in Fig. 6C. Following electrophoresis, Western analysis was performed to detect TBP. The position of TBP is shown. Mot1 and TBP were used at 50 nM, ATP was 100 M, and ADP-AlF 4 was used as described under "Materials and Methods." lead to instability of the native structure. A deeper understanding of the structural basis of the Mot1 defects reported here awaits future structural analysis.
Function of the Mot1 A Block-The A block is required for the Mot1⅐TBP interaction (Fig. 3A), yet most of the conserved polar residues of the A block can be changed to alanines without affecting cell viability (Table II). Only the mutation of a few of the first ten amino acids of the protein had any effect. In particular, mutation of Arg-7, Asp-9, and Arg-10 caused dominant inhibition of cell growth when the mutant gene was expressed from the GAL1 promoter (MOT1-101), and GAL1controlled alleles encoding mutations in Arg-7 or Asp-9 conferred temperature-sensitive growth. Overexpression of SPT15 rescued cells from overexpression of MOT1-101 (Fig. 3C), supporting the idea that the dominant negativity is due to an altered interaction with TBP. Mot1-101 protein is defective for binding TBP⅐DNA but has no obvious catalytic defect (Fig. 3D). One possibility is that the Mot1-101⅐TBP binary complex may be unusually stable: recycling of TBP after Mot1 action may be required in vivo. Formation of Mot1-101⅐TBP binary complexes could not be assessed in vitro using the pull-down assay, because purified Mot1-101 was found to interact nonspecifically with agarose beads (not shown), perhaps suggesting that the N terminus of Mot1-101 is not stably folded. Alternatively, the polar residues at the Mot1 N terminus may be important for an interaction with another protein that modulates the catalytic activity of Mot1 in vivo.
Two Putative Mot1-binding Sites on TBP-Previous results (19,38) and those in Fig. 5 demonstrate that the interaction of Mot1 with TBP requires lysine residues in TBP helix 2, on the convex surface of TBP opposite the DNA-binding site (Fig. 9A). These residues are required for both binary and ternary complex formation. A second putative Mot1-binding site is located on the concave DNA-binding surface of TBP and is defined by valine 71 and valine 161 (Fig. 9A). Mutation of these residues disrupts interaction with both DNA and Mot1, but these altered TBP molecules are unlikely to be simply misfolded, because they are expressed in soluble form at normal levels (not shown) and these same TBPs can stimulate transcription in vivo (51). A direct interaction between the DNA-binding surfaces of Mot1 and TBP can explain why the Mot1⅐TBP binary complex does not bind DNA. Interestingly, although Mot1 does not form a binary complex with TBP K127L, Mot1⅐TBP K127L⅐DNA ternary complexes were detectable, and these ternary complexes were disrupted in the presence of ATP (Fig. 5,  B and E). Similarly, a TBP with altered specificity for DNA binding supports Mot1⅐TBP⅐DNA ternary complex formation and ATP-dependent disruption (30) but is defective for binary interaction with BTAF1, a human Mot1 homolog (21). We suggest that Mot1 contacts the convex surface of TBP in both binary and ternary complexes and that Mot1 interaction with TBP alone requires a direct interaction between Mot1 and the TBP DNA-binding surface. In contrast, Mot1 does not directly interact with the TBP DNA-binding surface in Mot1⅐TBP⅐DNA ternary complex, but instead contacts the DNA upstream of the TATA box (13). Recent data (52) demonstrate that human TBP Lys-138 can affect DNA binding despite being located on the opposite side of the TBP DNA-binding surface. Therefore, an alternative possibility is that the effects of TBP DNA-binding surface mutations on Mot1 interaction result from reciprocal changes in TBP conformation rather than a direct interaction with Mot1. This possibility remains to be tested.
ATP Switches Mot1⅐TBP Binary Complex Affinity for DNA-The inability of Mot1⅐TBP binary complexes to bind DNA can be overcome by addition of ATP and the use of a DNA probe that is too short to allow Mot1 binding (13). We interpreted this result to indicate that the binary complex dissociates in the presence of ATP. On the other hand, an immobilized binary complex does not dissociate in the presence of ATP (19), suggesting that both DNA and ATP are required for dissociation. Here we report that ATP hydrolysis by the binary complex does not cause the binary complex to dissociate but, rather, that ATP induces a change in Mot1⅐TBP conformation that allows TBP to bind to DNA and Mot1 to be released. Experiments with the non-hydrolyzable ATP analog ADP-AlF 4 provided additional support for this model. Although binding of ADP-AlF 4 by Mot1 is not sufficient to drive Mot1⅐TBP⅐DNA ternary complex dissociation, ADP-AlF 4 did allow the binary complex to load onto DNA (Fig. 7A).
Mechanism of Disruption-Several mechanisms have been proposed for protein⅐DNA disruption by the Snf2/Swi2-related ATPases, including Mot1 (13,28,53,54). In contrast to mechanisms employed by at least some chromatin remodeling enzymes, Mot1 does not use ATP hydrolysis to propagate DNA bending, twisting, or strand separation through the TATA box (13). Mot1 also does not use ATP hydrolysis to track processively along DNA (30). Mot1-mediated changes in TBP confor- The data support a model in which the Mot1⅐TBP complex is stabilized by interactions between Mot1 and both the convex and concave surfaces of TBP, and the binary complex does not readily dissociate or bind DNA. An alternative possibility that is consistent with the data is that Mot1 binding induces a conformational change in TBP (shown as a distorted TBP in the Mot1⅐TBP binary complex), and the altered conformation of TBP binds DNA poorly. At one step in the ATP hydrolysis cycle, mimicked by binding of ADP-AlF 4 , the Mot1⅐TBP complex has an altered conformation in which the DNA-binding surface of TBP is either transiently accessible to DNA or TBP assumes its high affinity DNA binding conformation. ATP hydrolysis induces dissociation of the Mot1⅐TBP binary complex from DNA by weakening interaction of TBP with DNA, possibly through formation of contacts between Mot1 and the TBP DNA-binding surface. mation have been proposed to explain how Mot1 regulates the interaction between TBP and DNA (15). In support of the TBP conformational change model, human TBP⅐K138 modulates DNA binding affinity (52), a result suggesting that amino acids distal to the DNA-binding surface can affect DNA binding by directing a change in TBP conformation. Because this residue is critical for the interaction of yeast TBP and Mot1 (Fig. 5), it is possible that Mot1 interaction with the convex surface of TBP causes a change in the conformation of the TBP DNAbinding surface that is modulated by ATP.
The effects of TBP DNA-binding surface mutations on Mot1 interaction are most simply explained, however, by proposing a direct interaction between Mot1 and the DNA-binding surface of TBP. Combining this with previous observations, we propose the mechanism shown in Fig. 9B. The catalytic cycle then involves an ATP-driven insertion of the Mot1 N terminus into the TBP⅐DNA interface. This "power stroke" results in disruption of TBP⅐DNA contacts and the formation of new interactions between Mot1 and the DNA-binding surface of TBP. Once separated from DNA, the binary complex can hydrolyze ATP in a process that involves dramatic conformational changes in which the Mot1 N terminus alternates position in and out of contact with the DNA-binding surface of TBP. The conformation in which the DNA-binding surface of TBP is "open" can be trapped with ADP-AlF 4 , a state in which TATA-containing DNA can bind to TBP. Note that in the model (Fig. 9B), the conformation of TBP is different in the Mot1⅐TBP binary complex than when TBP is free, reflecting the possibility that Mot1 may induce a conformational change in TBP as part of its catalytic mechanism.
Mechanistic Insights Provided by Mot1-NC2 Interaction-Chromatin immunoprecipitation experiments established that both Mot1 and the NC2 subunit Bur6 are localized to the promoters that they regulate (43,55). Microarray experiments have shown that Mot1 and Bur6 regulate many of the same genes (43,45). The results in Fig. 8 demonstrate that Mot1 and NC2 can occupy the same promoter at the same time and show that NC2 does not impede the catalytic activity of Mot1. Thus, if such catalytic activity is altered at some promoters, as has been suggested to explain how Mot1 can activate the expression of some genes (43), this putative change in biochemical activity must depend on promoter-associated factors other than the NC2 complex. In addition, because Mot1 does not interfere with access of NC2 to TBP⅐DNA, Mot1 is unlikely to contact the underside of the TBP⅐DNA complex and the catalytic action of Mot1 is unlikely to require contact with the major groove of the TATA box.