Efficient binding of NC2.TATA-binding protein to DNA in the absence of TATA.

Negative cofactor 2 (NC2) forms a stable complex with TATA-binding protein (TBP) on promoters. This prevents the assembly of transcription factor (TF) IIA and TFIIB and leads to repression of RNA polymerase II transcription. Here we have revisited the interactions of NC2.TBP with DNA. We show that NC2.TBP complexes exhibit a significantly reduced preference for TATA box sequences compared with TBP and TBP.TFIIA complexes. In chromatin immunoprecipitations, NC2 is found on a variety of human TATA-containing and TATA-less promoters. Substantial amounts of NC2 are present in a complex with TBP in bulk chromatin. A complex of NC2.TBP displays a K(D) for DNA of approximately 2 x 10(-9) m for a 35-bp major late promoter oligonucleotide. While preferentially recognizing promoter-bound TBP, NC2 also accelerates TBP binding to promoters and stabilizes TBP.DNA complexes. Our data suggest that NC2 controls TBP binding and maintenance on DNA that is largely independent of a canonical TATA sequence.

Negative cofactor 2 (NC2) forms a stable complex with TATA-binding protein (TBP) on promoters. This prevents the assembly of transcription factor (TF) IIA and TFIIB and leads to repression of RNA polymerase II transcription. Here we have revisited the interactions of NC2⅐TBP with DNA. We show that NC2⅐TBP complexes exhibit a significantly reduced preference for TATA box sequences compared with TBP and TBP⅐TFIIA complexes. In chromatin immunoprecipitations, NC2 is found on a variety of human TATAcontaining and TATA-less promoters. Substantial amounts of NC2 are present in a complex with TBP in bulk chromatin. A complex of NC2⅐TBP displays a K D for DNA of ϳ2 ؋ 10 ؊9 M for a 35-bp major late promoter oligonucleotide. While preferentially recognizing promoter-bound TBP, NC2 also accelerates TBP binding to promoters and stabilizes TBP⅐DNA complexes. Our data suggest that NC2 controls TBP binding and maintenance on DNA that is largely independent of a canonical TATA sequence.
TATA-binding protein (TBP) 1 binds to eukaryotic promoters to nucleate initiation of transcription (1)(2)(3). The crystal structure of the TBP⅐DNA complex revealed a saddle-shaped conformation of TBP in which the highly conserved carboxyl terminus forms a concave surface that interacts with the minor groove of DNA. As a consequence, the minor groove becomes dramatically widened, and the DNA is severely bent (4). The characteristic bend as well as the TBP conformation is conserved in the co-crystal structures of TBP⅐TFIIA⅐DNA (5,6), TBP⅐TFIIB⅐DNA (7), and TBP⅐NC2⅐DNA (8).
The thermodynamics and kinetics of TBP-TATA interactions have been investigated in detail. Both yeast and human TBP show specificity for TATA boxes. Values for the K D are in the range of 5 ϫ 10 Ϫ10 to 2 ϫ 10 Ϫ8 M for yeast TBP (9 -13) and 4.6 ϫ 10 Ϫ10 to 1 ϫ 10 Ϫ9 M for human TBP (14,15). Variations may be accounted for in part by the specific conditions and the different methods employed, which are gel shifts and footprints versus solution FRET studies. Specificities of TBP for TATA range from Ͻ1 order of magnitude if a canonical TATA is compared with single point mutants in the TATA sequence to roughly 3 orders of magnitude relative to random DNA (14). Human TBP has been proposed to bind DNA through the conserved domain in the usual specific manner or, alternatively, in a nonspecific manner that depends on its non-conserved amino-terminal region (16).
Several factors modulate binding of TBP to DNA. The TBPassociated factors (TAFs) support binding of TBP specifically to TATA-less promoters via DNA contacts through cognate promoter sequences (17,18). The general transcription factor TFIIA accelerates and stabilizes binding of TBP to TATA boxes (19 -21). It also functions as a co-activator for some transcriptional activators (22)(23)(24)(25)(26). TFIIA further functions as an antagonist of NC2 (also called Dr1/DRAP) (27,28). Evidence for this equilibrium between NC2 and TFIIA stems from both biochemical investigations in human cells and genetic studies in yeast in which mutants in TFIIA genes were found to suppress an impaired NC2 gene (29 -32).
NC2 was originally identified in extracts from human cells as a factor that binds TBP and inhibits RNA polymerase II transcription in reconstituted systems (28,33,35,36). The cofactor is essential for the growth of yeast, and deletion of the gene encoding NC2␣ led to an early embryonic phenotype in mice (34). NC2 is a potent repressor of TBP-dependent RNA polymerase II transcription in human in vitro systems. Oxygen deficiency induces binding of NC2 to certain mammalian genes correlated to repression of transcription, which argues for a similar role in vivo (37). However, in Saccharomyces cerevisiae, NC2 seems to affect gene transcription both positively and negatively (38,39). Chromatin immunoprecipitation (ChIP) data suggested that the cofactor localizes to both inactive and active genes. The positive effect is not entirely understood at present, but it may reflect the capacity of NC2 to facilitate binding of TBP to promoters (40).
The NC2 complex consists of two subunits, named NC2␣ (Drap1) and NC2␤ (Dr1), which dimerize through histone fold domains of the H2A/H2B type present in the amino termini (8,27,41). Consistent with a critical role in transcription, the histone fold regions, together with two carboxyl-terminal helices in NC2␤, are highly conserved from yeast to man (27,29,(42)(43)(44)(45). Here we study the binding of NC2 to mammalian genes using gel shift assays and chromatin immunoprecipitations. NC2 is shown to localize to the promoter regions of many genes. Ratios of TBP to NC2 vary from one gene to another without uncovering an obvious correlation to the presence of bona fide TATA boxes. In fact, in vitro analyses revealed that NC2⅐TBP displays moderate sequence specificity for TATA. We further show that NC2 increases both the on-rate of TBP to DNA and the stability of TBP⅐DNA complexes.

Expression and Purification of Recombinant Proteins-Human NC2
subunits were expressed as both two single subunits and bicistronic co-expression constructs in pET11d as described previously (46). Truncation mutants were constructed using PCR primers harboring a stop codon at the desired position. All vectors were confirmed by sequencing. The human TBP, also in pET11d, was expressed and purified as described previously (47,48). Human TFIIA was expressed and purified as described elsewhere (31).
Longer DNA fragments used for EMSA (217 bp) were isolated from pB2-MLP carrying a 99-bp EcoRI/SmaI fragment of the AdML promoter (49). A TATA to TGTA mutation in this vector (pSO18) was produced with the QuikChange site-directed mutagenesis kit (Stratagene). EcoRI fragments of both vectors were purified and labeled with [␣-32 P]dATP.
Chromatin Immunoprecipitation-ChIP was performed with a rat monoclonal NC2␣ antibody (4G7) and polyclonal TBP antibody on formaldehyde cross-linked Jurkat cells. In brief, cells were fixed in 1% formaldehyde for 9 min at room temperature, sonicated, and run on a CsCl gradient as described previously (46). Antibody and extract were incubated overnight, followed by incubation with a mixture of protein A-and G-Sepharose beads (Amersham Biosciences). After extensive washing, complexes were eluted from the beads, cross-links were reversed, and DNA was purified by phenol-chloroform extraction and precipitated. Quantitative PCR was performed (28 cycles; sequences of primers are available upon request).
Mononucleosome Preparation-The chromatin pellet of extracted HeLa nuclei (36) was resuspended using a Dounce homogenizer in buffer (15 mM HEPES-KOH, pH 7.4, 15 mM NaCl, 60 mM KCl, 2 mM MgCl 2 , 0.2 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 2 g/ml pepstatin, and 3 g/ml aprotinin). The pellet was collected by centrifugation (200 ϫ g, 5 min) and washed twice with the same buffer. CaCl 2 was added to a final concentration of 0.23 mM. 200 units of micrococcal nuclease (N-5386; Sigma) were added and incubated at 37°C for 3-20 min. The reaction was stopped by the addition of EDTA to a final concentration of 5 mM, incubated on ice for 5 min, and centrifuged at 200 ϫ g for 15 min. The presence of mononucleosomes in the supernatant was confirmed by sizing columns, protein gels, and digestion with proteinase K and subsequent analysis of the DNA on agarose gels.
Immunoprecipitation-Protein G beads and NC2␣ monoclonal antibody (4G7) were incubated for at least 4 h. Mononucleosome fraction or nuclear extract was added; adjusted to BC150 supplemented with phenylmethylsulfonyl fluoride (final concentration, 5 mM), dithiothreitol (final concentration, 1 mM), and IGEPAL CA-630 (final concentration, 0.1%); and incubated for 3 h at 4°C with gentle mixing, followed by four rounds of washes with 1 ml of BC150 each. Beads were recovered by centrifugation for 3 min at 300 ϫ g. NC2 was eluted with SDS loading buffer (4 min at 95°C).
Calculation of Active Protein Concentration-Either NC2 or TBP was kept at limiting concentration while the other protein was in excess, and DNA was titrated until no further increase in complex formation was observed. This was defined as the active concentration, which is always referred to in the text. The active amount of TBP was ϳ65% in the specific preparation and ϳ60% for NC2 relative to total protein.
Calculation of the Equilibrium Dissociation Constant K D -The equation used to determine the binding constant derives from the equilibrium binding constant: is the concentration of NC2⅐TBP⅐DNA complex at equilibrium, and [protein] and [DNA] are the equilibrium concentrations of TBP and DNA, respectively. To simplify the equation, TBP⅐NC2 was treated as a single protein entity.

NC2⅐TBP Is Localized on TATA and TATA-less Promoters in
Vivo-ChIP experiments were performed to monitor NC2 and TBP on the promoters of endogenous human genes in Jurkat cells. NC2 was found on all promoters tested. NC2 and TBP levels are high and the NC2/TBP ratio is low at the highly active H2A/H2B locus. The HLAA locus carrying a cryptic TATA displayed the highest level of NC2. The TATA-less eIF4E promoter also binds NC2, whereas on the promoter of interleukin-2, TBP and NC2 levels are low, although it harbors a reasonable TATA box (Fig. 1). The limited set of data does not allow for a statistical correlation of NC2 occupancy relative to the core promoter structure. Presently, the data suggest that NC2 binds to many promoters in human cells. This binding does not require the presence of a TATA box. Finally, NC2/TBP ratios vary significantly from one promoter to another.
NC2 and TBP Are Present in Bulk Chromatin-Localization of NC2 on many genes may indicate that the factor resides in chromatin together with TBP. To test this hypothesis, mononu- cleosomes were prepared from chromatin extracts of HeLa cells and tested for the presence of NC2. Approximately 10 -20% of total NC2 was found in the mononucleosome fraction. NC2␣ was precipitated using a monoclonal antibody described previously (46). The precipitated material was analyzed for the presence of NC2␤, TBP, TAFs, and BTAF1 (formerly called TAF170/172) using polyclonal antibodies. TBP, but not TAFs or BTAF1 (50), co-precipitated with NC2 ( Fig. 2A). We also could not detect histones in the precipitated material (data not shown). Given the stringency of the immunoprecipitation, this does not exclude interactions of NC2 with BTAF1 and the TFIID complex. Nevertheless, the data argue for a strong preference of NC2 for free TBP. Notably, antibodies did not co-precipitate detectable amounts of TBP with NC2 in HeLa nuclear extracts. We reasoned that co-precipitation from chromatin extracts might be facilitated by DNA (Fig. 2B). To test this hypothesis, the immunoprecipitated material was treated with DNase I. This resulted in a complete loss of TBP in the NC2 immunoprecipitation (Fig.  2C), arguing for interaction of TBP and NC2 through DNA. It is somewhat puzzling that promoter and coding regions of the H2A/ H2B locus seemed underrepresented in NC2 immunoprecipitations compared with the input as analyzed by PCR (data not shown). This could be explained if TBP⅐NC2 localizes to open chromatin, in which the DNA is readily accessed by micrococcus nuclease. Although the average length of micrococcus nucleasetreated DNA was chosen to be in the range of kilobases in our analysis, hypersensitivity could perhaps lead to trimming of the genomic DNA flanking the TBP⅐NC2 binding sites, which could render them inaccessible to PCR analysis. This hypothetical model and the ChIP analysis support the notion that NC2 localizes to active promoters. However, at present, the exact nature of the preferred binding sites of chromatin-bound TBP⅐NC2 remains elusive.
Even if these were in part not promoter regions, formation of a TBP⅐NC2 could make sense in that it could prevent transcription from non-promoter regions. This primarily requires an excess of NC2 over TBP. Western analysis of Jurkat whole cell extracts revealed that NC2 is indeed present in moderate excess over TBP (with TBP comprising ϳ0.11 ng/g and NC2 comprising ϳ0.18 ng/g of the total protein mass; Fig. 2D). In addition, part of TBP is probably engaged in complexes, for example, with the class I and class III TAFs. Another major part is thought to be in a complex with BTAF1. In summary, NC2 is probably present in significant excess over TBP in the extracts of human cells.
The NC2⅐TBP Complex Shows Moderate Preference for TATA-EMSA was used to compare the binding of NC2⅐TBP to oligonucleotides comprising core regions of the selected promoters studied above in ChIP. Consistent with the in vivo localization, NC2⅐TBP bound to all promoters tested, irrespective of the presence of a consensus TATA motif. A moderate preference for promoters harboring a consensus TATA sequence (H2B and

NC2 Limits Specificity of TBP for TATA
interleukin-2) as compared with promoters with a poor TATA sequence (HLAA and V␤; Fig. 3A) was seen. In contrast to NC2⅐TBP, and as expected from the published data, TFIIA⅐TBP exhibits a clear preference for TATA (Figs. 3C and 4B). Specific TFIIA⅐TBP binding to TATA is seen even at high concentrations where NC2⅐TBP complexes do not show priority for TATA (Fig. 3, B and C). To document the specific influence of TATA, point mutations were analyzed in the context of the AdML promoter. Again, NC2⅐TBP complexes did not discriminate between wt and mutant oligonucleotides at nanomolar concentrations (Fig. 4A). However, a moderate preference for TATA (ϳ3-fold) was seen if limiting concentrations of TBP and NC2 were employed (Fig. 4C). Competition experiments further confirmed the limited specificity for TATA (Fig. 4D). To analyze the specificity of the NC2⅐TBP complex to DNA under more physiological conditions, the binding reaction was conducted at 140 mM potassium glutamate. Also, under these stringent conditions, TBP⅐NC2 binds DNA with very limited specificity for TATA (Fig. 4E).
The reduced specificity of TBP binding in the presence of NC2 was also seen with a 217-bp fragment comprising 80 bp of AdML promoter flanked by polylinker sequences and containing either a wt TATA sequence or a single point mutation (TGTAAAA). The mutation did not influence binding of the (fast) monomeric TBP⅐NC2 complexes (Fig. 4F). Under comparable conditions, the TFIIA⅐TBP complex that is seen on the wt fragment was not formed on a TGTA mutant. Further consistent with low specificity for TATA, NC2⅐TBP complexes usually form a ladder of monomeric, dimeric, and trimeric complexes (and so forth) on longer DNA fragments that contain both TBP and NC2 (data not shown). 2 The appearance of higher order complexes in the case of TBP⅐TFIIA probably results from binding to TATA-like sequences present in the polylinker. This is not seen if limiting concentrations of TBP are used and, contrary to TBP-NC2, is not seen on other DNA fragments.
Efficient Recognition of DNA-bound TBP by NC2-Next we wished to determine the binding constant of NC2⅐TBP⅐DNA complexes. Initially, the limiting factor for complex formation was identified via titration of TBP and NC2 at constant and 2 G. Stelzer, E. Piaia, and M. Meisterernst, manuscript in preparation. limiting (0.1 nM) DNA concentrations (Fig. 5A). At 0.5 nM TBP and high (20 nM) NC2, templates were not yet saturated (note that TBP was present in excess over DNA). If TBP levels are raised to 3 nM, low concentrations of NC2 (0.15 nM) shift more than 50% of the promoter. We conclude that TBP, but not NC2, is limiting for complex formation. Similar results were obtained at somewhat higher (0.5 nM) DNA concentrations (Fig. 5B). This set of experiments allows for a rough estimation of the affinity of NC2 for TBP⅐DNA (K D 1-2 ϫ 10 Ϫ10 M). As demonstrated below and consistent with the notion that TBP is the limiting factor in complex formation, the dissociation constant is significantly below the K D of TBP⅐NC2 for DNA. To deter-mine the K D of TBP⅐NC2 for DNA, the DNA was titrated, whereas TBP and NC2 concentrations were kept constant. To accurately determine the K D , all concentrations must be kept in the expected (nanomolar) range of the equilibrium constant. Active concentrations were used for calculation of the dissociation constant. A K D of 2.4 ϫ 10 Ϫ9 M was determined on the 35-bp AdML promoter oligonucleotide (Fig. 6). The experiment was repeated several times with a similar outcome.
To assess the limitation of this EMSA-based method for determination of binding constants, we next asked whether TBP and TBP⅐NC2 complexes dissociate in the gel from DNA during electrophoresis. Complex levels were measured 50 and 110 min, respectively, after starting electrophoresis. Within 1 h, 50% of the TBP but only 20% of the TBP⅐NC2 was dissociated (Fig. 7A). We conclude that TBP⅐DNA complexes are less stable than TBP⅐NC2⅐DNA complexes, at least under these experimental conditions. The relative stability of TBP⅐NC2 suggests that the K D values may represent a reasonable approximation. However, taking into account that electrophoresis takes several hours, binding constants will nevertheless represent a lower estimate.
NC2 Accelerates Binding of TBP to DNA-The on-rate of TBP and TBP⅐NC2 for DNA was analyzed in EMSA. Complex formation was measured after different incubation times (Fig.  7B). In the case of TBP alone, 50% saturation of the templates was reached after 30 min, whereas it took only 2 min for TBP⅐NC2 to reach this level of binding. Whereas the absolute numbers of the on-rates represent rough estimations, the data clearly show that NC2 markedly (ϳ10-fold) increases the onrate of TBP for DNA.
The Carboxyl-terminal Domain of NC2␣ Increases Affinity for TBP⅐DNA-Carboxyl-terminal regions of NC2 subunits are not required for dimerization, but they stimulate TBP⅐NC2⅐DNA complex formation. There is evidence that the carboxyl termini of both subunits contribute to repression of transcription (27). This is ex-plained by the x-ray structure, which showed direct binding to TBP (8). Here we asked whether the carboxyl-terminal regions affect (i) affinity and (ii) specificity for TATA. Two truncated forms of NC2␣ (including amino acids 1-132 and 1-157) were generated and tested in EMSA on a wt and a mutant AdML core promoter, in combination with wt NC2␤ and a truncated NC2␤1-112 lacking the TBP interaction domain (Fig. 8). Deletion of the carboxyl-terminal residues had little impact (compare NC2␣157 with NC2␣wt comprising 205 amino acids), but further deletion of 25 amino acids (NC2␣132) drastically reduced affinity. The affinity was further reduced if NC2␤112 was used instead of NC2␤wt (comprising 176 amino acids). The combination of the two short versions (NC2␣132 and NC2␤112) binds TBP⅐DNA poorly. Thus, human NC2␣ carboxyl-terminal regions and, specifically, the negatively charged region between amino acids 132 and 157 contribute to complex formation. However, the carboxyl-terminal regions do not significantly modulate specificity for TATA. Notably, up to a 5-fold preference for binding to TATA was measured here. This is ϳ2-fold better than the specificity documented above. The reason for the increase is the use of a TGCTGGG mutation in TATAAAA, whereas single and double point mutations were analyzed above. Thus, these data further underline the notion of moderate specificity of TBP⅐NC2 for TATA.

DISCUSSION
Here we have analyzed the binding of NC2⅐TBP complexes to core promoters both in vitro and in vivo. Chromatin immunoprecipitation experiments indicated binding of NC2 to all human promoters tested. The presence of NC2 was apparently not correlated to the presence of TATA boxes. Consistent with a general role on mammalian genes, a substantial fraction of NC2 was found in a complex with TBP and DNA in bulk chromatin. Kinetic and thermodynamic data showed that NC2 (i) accelerates binding of TBP to DNA and (ii) stabilizes TBP on DNA. Surprisingly, TBP⅐NC2 exhibited little specificity (Ͻ1 order of magnitude) for TATA boxes compared with random DNA. These data indirectly point to a critical role of NC2 on TATA-less promoters. On these, TBP alone cannot recognize core regions following the opening of the chromatin structure. Activators, general transcription factors, and TAFs are thought to escort and bind TBP to core regions. NC2, in turn, has the capacity to both support the binding and maintain TBP at these promoters. If NC2 could be removed in cells, it might ultimately support productive initiation of transcription. This model could provide a tentative explanation for the positive effects seen on many yeast genes (39,51).
A K D of 2.4 nM for binding of TBP⅐NC2 to DNA has been determined. These affinities are in the lower range of the published K D for TBP alone. This is explained in part by the gel method employed here, which usually leads to a lower estimate of the affinity (and, conversely, a higher estimate of the corresponding dissociation constant). As one example, we showed that TBP dissociates during electrophoresis, whereas TBP⅐NC2 complexes seem to be more stable. Another potential source of errors is the salt concentrations, which are often kept far below physiological levels in vitro. Monovalent anions such as Cl Ϫ are rare in cells, in which negative charges are brought about mainly by nucleic acids, phosphates, and proteins. We have begun to analyze the impact of more physiological salt concentrations on TBP⅐NC2 DNA interactions. Again, TBP proved to be more sensitive than TBP⅐NC2, which in turn bound DNA with high affinity even at 140 mM potassium glutamate. Nonetheless, the TBP⅐NC2⅐DNA binding constant also predictably represents a lower estimate. This is corroborated by preliminary evidence that argues for enhanced stability of TBP⅐NC2 complexes on longer DNA, whereas a short 35-bp core promoter fragment was used in this study.
DNA-bound TBP Is Efficiently Tracked by NC2-This is concluded from the K D of NC2 for a preformed TBP⅐DNA complex, which is in the range of 0.1 nM. This is approximately 1 order of magnitude below the binding constant of TBP⅐NC2 for DNA. These thermodynamic data lead to the prediction that NC2 will form a complex with TBP also if it is erroneously bound to open chromatin regions. Indeed, the nonspecific binding capacity of NC2⅐TBP also allows for interaction with DNA outside of promoter regions. This might be specifically relevant to metazoans, in which genes are large and activator-binding sites are found throughout the genome. The functional consequence of tracking TBP outside of promoter regions is of specific interest: in this situation, NC2 could prevent transcription from nonpromoter regions.
Contrary to NC2, TBP⅐TFIIA displays a much reduced affinity for promoters carrying mutations in or lacking TATA. This is entirely consistent with previous investigations (19,20) showing reduced affinity and stability of TBP for TATA mutants compared with TATA (21). Thus, our data predict that TFIIA is more important on TATA-containing promoters than on TATA-less promoters. In human cells, a genome-wide ChIP analysis has not been conducted. In yeast, TFIIA, like NC2, is essential for growth (31). On the other hand, yeast is viable at low TFIIA levels. Under these conditions, TFIIA inactivation reduces transcription from many yeast genes. The reduction is 2-3-fold; however, a preference for TATA has not been specifically noted (52). Given, however, that (i) both TBP and TFIIA localization are influenced by regulatory surfaces as well as by contacts with general transcription factors and (ii) ChIP experiments monitor stable states (53), the preference of TFIIA for defined core promoter structures may not be readily deduced from array data. We have confirmed the results with yeast TBP, but yeast TFIIA remains to be analyzed. This is of potential interest, given that core promoter architecture differs substantially from yeast to man.
Our earlier data suggested that TFIIA competes with NC2 on promoters (notably, these were always TATA-containing promoters). Although a comparison of TFIIA and NC2 in the corresponding co-crystals shows little overlap (8,54), the fac-tors compete for TBP in biochemical DNA binding experiments as well as in transcription (55). TFIIA⅐TBP complexes bind TATA with a K D of ϳ0.02 nM (56), explaining in general the capacity of TFIIA to compete with NC2 for binding to TBP bound to TATA. Also, the genetic selection of a suppressor of NC2 argued for an equilibrium between NC2 and TFIIA. One of the suppressors proved to encode a point mutation in one TFIIA subunit (TOA1 V251F) that was shown to render the TFIIA⅐TBP⅐DNA complex less stable (31).
Inefficient binding of TFIIA to TBP on non-TATA sequences raises the question as to how NC2 can be removed from TBP on these sites. The biochemical analysis as well as the x-ray structure suggests this to be a prerequisite for productive initiation of transcription. One candidate factor that may play a role independent of a TATA sequence is BTAF1 (Mot1). Gumbs et al. (13) reported that BTAF1, like NC2, drastically diminishes the preference of TBP for TATA. Also, in the case of B-TFIID (TBP in association with BTAF1), the affinity for non-TATA DNA sequences seems to be enhanced, rather than the affinity for TATA being reduced (13). Further reminiscent of NC2, FIG. 7. A, NC2⅐TBP⅐DNA complex is more stable than the TBP⅐DNA complex in native gels. EMSA with 35-bp AdML promoter samples divided between two gels that were run in parallel for 50 or 110 min. 28.6 nM TBP, 2.2 nM NC2, and 2.5 nM DNA were used. The y axis gives the percentage of complex formation compared with the total DNA (n ϭ 3; results are the mean Ϯ S.E.). B, NC2 accelerates binding of TBP to DNA. Samples were incubated for increasing time periods (1,3,10,20,40, and 50 min), 2.5 nM 35-bp AdML promoter. The first half of the gel contains TBP (32 nM), and the second half contains TBP and NC2 (8 and 2.2 nM, respectively). The gel was started running as the first sample was added (only TBP DNA at time 0) and stopped briefly at the time points when the additional samples were added. The gel was run for 50 min after loading of the last sample. BTAF1 has been proposed to affect transcription both positively and negatively, depending on the gene context. Moreover, in yeast, the two factors were found to localize on promoters in a mutually exclusive manner (57). Consistent with this notion, NC2 and BTAF1 were shown to compete for binding to TBP on DNA in vitro. Beyond this, it has been postulated that BTAF1 may have the capacity to disrupt NC2 complexes (58). It will be of considerable interest to understand in more detail how the assembly and disassembly of NC2⅐TBP complexes are regulated in vivo. FIG. 8. Deletion derivatives of NC2 lacking the carboxyl termini show reduced affinity and moderate specificity for AdML promoter TATA. Data are based on EMSA of TBP⅐NC2 derivatives on the standard 35-bp AdML promoter harboring a consensus TATA sequence or a mutation to TGCTGGG at limiting conditions. Numbers refer to the percentage of complex formation relative to the total amount of DNA. The truncated forms of NC2␣ used were ␣1-157 or ␣1-132. NC2␤ was either full length or ␤1-112. The concentrations were 0.5 nM DNA, 1.5 nM TBP, 0.5 nM NC2␤ subunit, and 2 nM NC2␣.