HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 8, 5841-5848, February 22, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/8/5841    most recent M103651200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Frontini, M. Articles by Mantovani, R. Search for Related Content PubMed PubMed Citation Articles by Frontini, M. Articles by Mantovani, R.

NF-Y Recruitment of TFIID, Multiple Interactions with Histone Fold TAF_IIs^*

Mattia Frontini, Carol Imbriano, Alberto diSilvio, Brendan Bell^§, Alessia Bogni, Christophe Romier^§, Dino Moras^§, Laszlo Tora^§, Irwin Davidson^§, and Roberto Mantovani^¶

From the  Dipartimento di Biologia Animale, Università di Modena e Reggio, Via Campi 213/d, Modena 41100, Italy and the ^§ Institut de Genetique et de Biologie Moleculaire et Cellulaire. 1, Rue L. Fries, BP163, Illkirch C.U. Strasbourg 67404, France

Received for publication, April 24, 2001, and in revised form, September 26, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The nuclear factor y (NF-Y) trimer and TFIID contain histone fold subunits, and their binding to the CCAAT and Initiator elements of the major histocompatibility complex class II Ea promoter is required for transcriptional activation. Using agarose-electrophoretic mobility shift assay we found that NF-Y increases the affinity of holo-TFIID for Ea in a CCAAT- and Inr-dependent manner. We began to dissect the interplay between NF-Y- and TBP-associated factors PO1II (TAF_IIs)-containing histone fold domains in protein-protein interactions and transfections. hTAF_II20, hTAF_II28, and hTAF_II18-hTAF_II28 bind to the NF-Y B-NF-YC histone fold dimer; hTAF_II80 and hTAF_II31-hTAF_II80 interact with the trimer but not with the NF-YB-NF-YC dimer. The histone fold 2 helix of hTAF_II80 is not required for NF-Y association, as determined by interactions with the naturally occurring splice variant hTAF_II80. Expression of hTAF_II28 and hTAF_II18 in mouse cells significantly and specifically reduced NF-Y activation in GAL4-based experiments, whereas hTAF_II20 and hTAF_II135 increased it. These results indicate that NF-Y (i) recruits purified holo-TFIID in vitro and (ii) can associate multiple TAF_IIs, potentially accommodating different core promoter architectures.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Gene expression is regulated by promoter and enhancer elements recognized by gene-specific DNA-binding proteins and by general transcription factors (1). At a higher level, it is controlled by chromatin structures, whose fundamental unit is the nucleosome, a complex formed by core histones H2A, H2B, H3, H4, which wrap around them 146 base pairs of DNA (2, 3). Histones all share a conserved 65-amino acid histone fold motif (HFM)¹ that has low sequence identity but high structural resemblance (4). Crystallographic analysis showed that this motif is composed of three/four -helices separated by short loops/strand regions; this structure enables histones to dimerize and form non-sequence-specific interactions with DNA (5). Proteins containing the HFM are also involved in the basic mechanisms of transcription: (i) the two subunits of the TBP-binding NC2, also called Dr1/DRAP1, a global repressor of basal transcription (6, 7); (ii) some of the TBP-associated factors that are part of the TFIID, P/CAF, STAGA, and TFTC complexes (8-17); (iii) one subunit of the P/CAF complex (18); and (iv) two subunits of the CCAAT-binding activator NF-Y (19-21).

TFIID is a general transcription complex composed of TBP, responsible for TATA recognition, and of several associated factors, TAF_IIs, that constitute a link between gene-specific upstream activators and the general transcription machinery by recognizing TATA and/or initiator elements (reviewed in Refs. 22, 23). Some of the TAF_IIs appear to be present in specific sub-complexes of TFIID (24-33); 10-12 highly conserved subunits have been identified in yeast, Drosophila, and human and biochemically characterized in protein-binding assays and functional in vitro transcription experiments. Based on sequence homology, structure-function analysis, and crystallographic studies, hTAF_II80/dTAF_II60, hTAF_II31/dTAF_II40, hTAF_II28/dTAF_II27, and hTAF_II18/dTAF_II30 have histone-like structures (13, 15); hTAF_II20/hTAF_II135 and hTAF_II30 have also been included in this class (10, 11, 16, 17). Interestingly, gene inactivation of HFM-containing TAF_IIs in yeast implicates them in a rather broad, if not universal, role in transcriptional activation (34-38). This is unlike other TAF_IIs, whose inactivation in yeast suggests a more selective role in certain promoters (38-40).

Another protein containing HFMs is NF-Y, also termed CBF, the ubiquitous trimeric protein binding to the widespread CCAAT-box promoter element; it is composed of NF-YA, NF-YB, and NF-YC, all necessary for subunit association and DNA binding (reviewed in refs. 19 and 21). The H2B-H2A-like NF-YB-NF-YC subunits dimerize tightly via their HFMs, forming a complex surface necessary for NF-YA association. The resulting trimer has a high affinity and sequence specificity for the CCAAT sequence. Several types of indications link TFIID to NF-Y. First, their binding sites are either ubiquitous, in the case of TFIID, or quite frequent, because 25% of the promoters have NF-Y sites; both are found at highly conserved positions within a prototypical promoter (20). Second, biochemical evidence of direct interactions has emerged: (i) NF-YB and NF-YC bind to TBP through short subdomains within the larger yeast/human conserved parts and short basic residues in the HS2 of TBP (41); (ii) NF-YB is immunoprecipitated with an anti-TAF_II100 antibody from crude nuclear extracts and is present in immunopurified TFIID fractions and in high molecular weight complexes in glycerol gradient experiments, indeed suggesting association with additional proteins; (iii) the Q-rich regions of NF-YA and NF-YC interact with dTAF_II110 in vitro (42), a result consistent with the idea that Q-rich activators such as SP1 and cAMP-response element-binding protein function by binding to dTAF_II110-hTAF_II135 (43-48). Third, the lack of NF-Y binding has been associated with a closed chromatin configuration of the Xenopus HSP70 TATA-box region (49) and, very recently, with the inability to recruit TBP-TFIIB on the -globin promoter in vivo (50): These results clearly imply that NF-Y binding is essential for TFIID recruitment.

To study the NF-Y-TFIID connections, we employed the mouse MHC class II Ea promoter system (51, 52). Ea is a tissue-specific promoter active in B lymphocytes and other professional antigen-presenting cells (53). Like all other MHC class II promoters, it also requires the ubiquitous trimer RFX and lacks a functional TATA-box. We showed that TFIID binding to the Ea Inr is necessary for function in an in vitro transcription system. In this study, we began to dissect NF-Y-TFIID interplay with purified holo-TFIID, recombinant NF-Y, and isolated TAF_IIs.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Production and Purification of Recombinant NF-Y, hTAF_II31, hTAF_II80, hTAF_II28, hTAF_II18, hTAF_II20, hTAF_II1351-372, and holo-TF_IID-- Production and purification of recombinant NF-Y trimer were as described before, using wt NF-YA, wt NF-YB, and a TRX-His-NF-YC fusion protein (54). Recombinant His-tagged hTAF_II28 and hTAF_II18, GST-hTAF_II28 (9), GST-hTAF_II1351-372 (12), and GST-hTAF_II20 (9) were produced in Escherichia coli as soluble proteins and purified according to standard procedures. hTAF_II31, hTAF_II80, and hTAF_II80 were produced by the baculovirus expression system, either as single subunits or as a dimer, using standard protocols and purified to homogeneity. hTAF_II80 contains a FLAG tag. Holo-TFIID was immunopurified from HeLa cells with an anti-TBP antibody as previously detailed (9, 26, 27).

EMSA Analysis-- EMSAs of TFIID in agarose gels were as described in Ref. 52: holo-TFIID fractions were incubated in NF-Y buffer (20 mM HEPES, pH 7.9, 50 mM NaCl, 5% glycerol, 5 mM MgCl₂, 1 mM dithiothreitol) together with 10,000 cpm of ³²P-labeled Ea fragments; the total volume was 10 µl. After incubation for 45 min at 30 °C, we added 2 µl of 1× buffer containing bromphenol blue, and samples were loaded on a 1.5% agarose gel (Bio-Rad Ultrapure) in 0.5× TBE. Gels were run at 140 V for 90 min at 4 °C, transferred onto DE81 paper, vacuum-dried, and exposed. Three independent preparations of purified TFIID were used in EMSAs. The Ea fragments used in EMSA analysis were obtained by PCR and contained sequences from 115 to +60 of the Ea promoter, either wt, mutated in the Y box (Ls17 (51)), or in the Inr (Ls21 (52)).

Antibodies and Supershift EMSA-- For supershift experiments, anti-NF-YA and -NF-YB antibodies were purified on antigen columns (55). Monoclonal antibodies against TAF_IIs were as follows: 24TA and 26TA, anti-hTAF_II80; 22TA, anti-hTAF_II20; 15TA, anti-hTAF_II28; and 16TA, anti-hTAF_II18 (9, 12, 26). Monoclonal antibodies were purified by caprylic acid precipitation of ascites fluid followed by precipitation with 50% ammonium sulfate, resuspension in phosphate-buffered saline, and dialysis against NDB100 (100 mM KCl, 20 mM HEPES, pH 7.9, 20% glycerol, 0.5 mM EDTA). The hTAF_II31 rabbit polyclonal was a kind gift of Dr. A. Levine. Supershift experiments were performed by preincubating TFIID, with or without NF-Y, with the indicated antibodies (200 ng of purified monoclonal antibodies, 0.3 µl of the anti-hTAF_II31 polyclonal, 200 ng of purified anti-NF-YA and anti-YB) for 2 h on ice, before addition of the labeled DNA and further incubation at 30 °C for 45 min.

Protein-Protein Interactions-- Interactions with NTA-agarose columns were performed by incubating either crude bacterial extracts or purified His-tagged proteins (1 or 2 µg) in BC100 (100 mM KCl, 20 mM HEPES, pH 7.9, 10% glycerol, 5 mM imidazole, 5 mM -mercaptoethanol) with NTA-agarose (100-200 µl); the column was washed with BC300 and BC1000, containing 300 mM and 1 M KCl, respectively; proteins were eluted in BC100 buffer containing 300 mM imidazole, dialyzed against BC100, and assayed in Western blots. Immunoprecipitations were performed as follows: The NF-Y trimer (500 ng) and an equivalent amount of the indicated TAF_IIs were added to 25 µl of Protein G-Sepharose to which 7.5 µg of the purified anti-NF-YA7 monoclonal antibody had been previously bound. Incubation was pursued for 2 h on ice, unbound material was recovered after centrifugation, and the beads were washed with NDB100 with the addition of 0.1% Nonidet P-40. SDS buffer was added, and the samples were boiled at 90 °C for 5 min and loaded onto SDS gels. Western blots were performed according to standard procedures with the indicated primary antibody and a Pierce peroxidase secondary antibody. For multiple interactions, the filter was stripped, blocked with nonfat dry milk, and re-hybridized.

Transfections-- The eukaryotic expression vectors for NF-Y, hTAF_II28, hTAF_II18, hTAF_II20, hTAF_II135, and hTAF_II80 were described before (9, 12, 56). From PCR-hTAF_II31 (kindly donated by Dr. Levine) an EcoRI insert was cloned into pGAL4poly in-frame with the Gal4 DNA-binding domain, and the latter was excised by cutting with XhoI and a partial EcoRI digest. Mouse NIH-3T3 fibroblasts were co-transfected with 1-3 µg of activating plasmids, 2 µg of the plasmids containing the Luciferase reporter gene, and 3 µg of pNGal plasmid for control of transfection efficiency. The total amount of DNA was kept constant (at 15 µg) with Bluescript. All plasmids were purified by centrifugation using cesium chloride gradients. Cells were transfected with the standard calcium-phosphate method, recovered 48 h after transfection, washed in phosphate-buffered saline (150 mM NaCl, 10 mM sodium phosphate, pH 7.4), and resuspended in the Reporter assay reagent (Promega). Luciferase and -galactosidase activity were measured according standard procedures. A minimum of three independent transfections in duplicate was done; most of the values are based on 8-12 transfections.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Binding of NF-Y and TFIID to the Ea Promoter-- We have previously shown that the MHC class II Ea promoter is crucially dependent on a binding of holo-TFIID to a TdT-like initiator (52). We investigated the interactions between NF-Y and holo-TFIID in such a system using recombinant NF-Y and immunopurified holo-TFIID in agarose-EMSA. Fig. 1 (A and B) shows a dose response of TFIID either alone (lanes 5-7), or with a fixed amount of NF-Y (lanes 2-4): in Fig. 1A the NF-Y dose saturated the labeled fragment (lane 1), whereas in Fig. 1B lower doses of NF-Y were used. Two major complexes of different mobility, termed IIDa and IIDb, were generated by TFIID; when incubated with NF-Y, both complexes generated dissimilar electrophoretic mobilities, clearly arising as a result of co-incubation of NF-Y and TFIID (Fig. 1A, compare lanes 2-4 with 5-7; Fig. 1B, compare lanes 1-3 with 5-7): One complex migrated slightly more slowly than the NF-Y band, and another was further retarded and visible at higher TFIID concentrations. Note that the upper Y/IID complexes were visible at lower TFIID concentrations (Fig. 1A, compare lanes 3 and 6). As a control for the specificity of the interactions, we used an identical Ea fragment carrying a 10-bp mutation in the Y-box, known to abolish NF-Y binding, as well as in vivo and in vitro transcriptional activity of the promoter (51): As expected, binding of NF-Y, but not TFIID, was abolished, and the upper bands resulting from simultaneous interactions were not detected (Fig. 1C).

View larger version (58K): [in this window] [in a new window]   Fig. 1.   EMSA analysis of NF-Y and holo-TFIID binding to Ea core sequences. A, dose-response analysis of holo-TFIID in the presence (lanes 2-4) or absence of 1 ng of NF-Y (lanes 2-4, 0.2, 1, 3 µl, respectively); in lane 1, 1 ng of NF-Y was used. B, same as in A, except that a low amount of NF-Y (0.1 ng) was used in each of lanes 5-8. No protein was added in lane 1. C, dose response of NF-Y (0.1 ng in lanes 2, 5, 8, 11; 0.3 ng in lanes 4, 6, 9, 12) with (lanes 4-6, 10-12) or without (lanes 1-3, 7-9) 1 µl of TFIID. The wt Ea 90/+60 promoter fragment was used in lanes 1-6, whereas in lanes 7-12 we used a fragment of identical length containing a 10-bp mutation in the CCAAT box (LS17 (see Ref. 51)). D, dose response of NF-Y (0.01, 0.1, 1 ng, lanes 2-4 and 5-7, respectively) were incubated alone (lanes 2-4) or with 3 µl of TFIID (lanes 6-8). E, same as in D, except that 0.3 µl of TFIID was used. F, dose response of NF-Y (0.1 ng in lanes 2, 5, 8, 11; 0.3 ng in lanes 4, 6, 9, 12) with (lanes 4-6, 10-12) or without (lanes 1-3, 7-9) 0.3 µl of TFIID. In lanes 1-6 we used the wt Ea 90/+60 promoter fragment; in lanes 7-12 a fragment of identical length containing a 10-bp mutation in the Initiator (LS21 (see Ref. 52)) was used. G, antibody supershift of NF-Y/TFIID, TFIID, and NF-Y complexes: the indicated antibodies (anti-hTAFII80 monoclonals were 24TA in lanes 6 and 26TA in lanes 7) were incubated with TFIID on ice before addition of NF-Y (upper panel), with TFIID alone (middle panel), and with NF-Y (lower panel). For details of the antibodies, see "Materials and Methods." The control antibody was an anti-GATA1 monoclonal. H, supershift experiments as in C, with anti-NF-YB (lane 2)- and anti-NF-YA (lane 3)-purified polyclonal antibodies.

To verify the effect of TFIID on NF-Y binding, a reciprocal experiment was also performed, namely a dose-response analysis of NF-Y alone, or with two TFIID concentrations (Fig. 1, D and E, lanes 2-7). The TFIID pattern at high concentrations was clearly modified in a dose-dependent manner by NF-Y (Fig. 1D, compare lanes 1 with 2-4). Interestingly, the lowest amount of NF-Y employed (0.1 ng), which was barely sufficient to generate a visible band, modified the pattern of the TFIIDa and -b bands. When incubated with low amounts of TFIID, insufficient to shift Ea DNA, the NF-Y band was evident at lower concentrations (Fig. 1E, compare lanes 2 and 3 with 6 and 7). In experiments conceptually similar to those of Fig. 1C, we employed a fragment containing a 10-bp mutation in the Ea Inr region, known to cripple TFIID binding and Ea promoter function in vitro (52): The NF-Y complex, but not the TFIID or the Y/IID complexes, was generated by co-incubation of the two proteins (Fig. 1F), indicating that an intact Inr is required for the formation of the NF-Y-TFIID complexes.

Next, we wished to determine whether the Y/IID complexes observed in our EMSAs truly contain TAF_IIs. To this aim, we used antibodies specific for different TAF_IIs in supershift experiments. Fig. 1G indicates that anti-hTAF_II31, anti-hTAF_II30, anti-hTAF_II20, and two different anti-hTAF_II80 antibodies all modify the Y/IID complexes, whereas an irrelevant anti-GATA1 antibody had no effect (Fig. 1G, upper panel, compare lanes 1 and 2 with 3-7). In parallel, when challenged with the IIDa complex, these antibodies all showed interactions (Fig. 1G, middle panel), whereas none recognized the NF-Y complex (Fig. 1G, lower panel). Note that the anti-hTAF_II31 antibody, rather than supershifting the IIDa or Y/IID complexes, apparently inhibited binding of TFIID to DNA. Finally, we verified whether the Y/IID complexes also contain NF-Y, by challenging them with anti-NF-YA and anti-NF-YB antibodies. Indeed, both antibodies modified the mobilities of the complexes (Fig. 1H). The experiments shown in Fig. 1 (A-E) also suggest that the presence of NF-Y improves the binding capacity of holo-TFIID. We verified whether NF-Y can facilitate binding of TFIID by performing on-rate experiments: Binding of TFIID to DNA, in fact, is known to be a slow process, and the on-rates even on high affinity TATA-Inr elements are on the order of 20-40 min (Ref. 57, and references therein). On the contrary, NF-Y binding under our experimental conditions is extremely rapid, being completed after 1-2 min (Fig. 2A; see also Ref. 58). We incubated suboptimal amounts of TFIID (see Fig. 1E) with and without saturating amounts of NF-Y: A weak IID band was seen only after 30 min of incubation in the absence of NF-Y (Fig. 2B, lanes 6-9), whereas the upper Y/IID complexes were evident already after 2 min of incubation at 30 °C, and maximal after 5 min (Fig. 2B, lanes 2 and 3), strongly suggesting that NF-Y-CCAAT complexes recruit TFIID onto the Ea initiator.

View larger version (73K): [in this window] [in a new window]   Fig. 2.   On-rates of TFIID and NF-Y on the Ea promoter. A, EMSA on-rate experiments of NF-Y (1 ng). B, same as in A, except that NF-Y and TFIID (0.1 µl) were used. Experiments were performed by preincubating NF-Y alone (lane 1) for 15 min at RT and then adding holo-TFIID for the indicated time before loading a running agarose gel (lanes 2-5). Incubations of TFIID alone for the corresponding times are shown in lanes 6-9.

From this set of in vitro experiments with purified holo-TFIID and recombinant NF-Y, we conclude that: (i) complexes of different mobilities are formed upon simultaneous binding of TFIID and NF-Y to the Ea promoter; (ii) these complexes contain both TAF_IIs and NF-Y; (iii) binding of NF-Y to the CCAAT box and of TFIID to the initiator is required; and (iv) TFIID binding is remarkably facilitated when NF-Y is bound to DNA.

Binding of NF-YB-NF-YC to Histone Fold TAF_IIs-- A large number of subunits are present in holo-TFIID, and full reconstitution of the holo-TFIID complex with recombinant proteins has not been achieved yet. We therefore decided to dissect NF-Y-TFIID interactions by taking a reductionist approach, investigating the interactions between isolated HFM subunits of the two complexes. Recombinant proteins were produced in E. coli or Baculovirus and purified (Fig. 3). Note that the GST-hTAF_II1351-372 protein used hereafter is a mutant containing the C-terminal region with the HFM (16) but lacking the N-terminal 372 Q-rich region that mediates binding to Q-rich activators, such as SP1 and cAMP-response element-binding protein (46), and possibly contact NF-Y (42). The recombinant NF-YB-NF-YC dimer, containing the histone folds, was incubated with bacterial extracts containing GST-hTAF_II28, GST-hTAF_II1351-372, GST-hTAF_II20, or GST-hTAF_II20-GST-hTAF_II1351-372, and Sf9 extracts containing hTAF_II80-hTAF_II31. The complexes were purified over a nickel NTA-agarose column, exploiting the presence of His tags on NF-YC. Columns were washed with buffers containing 0.3 M and 1 M KCl, and eluted with 0.3 M imidazole. Flow-through, wash, and bound material were checked in Western blots with the respective antibodies. Results of the experiments are shown in Fig. 4. As expected, NF-YB was efficiently bound to the column, despite the lack of His tags (Fig. 4, upper panels). GST-hTAF_II20 and GST-hTAF_II28 were efficiently retained on the columns but were in the FT fraction in the absence of NF-YB-NF-YC, thus ruling out that any of the HFM proteins tested had any intrinsic affinity for NTA-agarose (Fig. 4 and data not shown). On the other hand, GST-hTAF_II1351-372, GST-hTAF_II20-GST-hTAF_II1351-372, and hTAF_II80-hTAF_II31 were not retained by the affinity column. The reverse approach was also tested, namely His-tagged hTAF_II28, hTAF_II18, and hTAF_II18-hTAF_II28 were incubated with an NF-YB-NF-YC5 dimer lacking His tags. In this experiment, we used an NF-YC mutant, YC5, that contains only the evolutionarily conserved part of NF-YC, fully capable to associate NF-YB. The NF-Y HFM subunits were found in the bound fractions with hTAF_II28 and hTAF_II18-hTAF_II28 but not with hTAF_II18.

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Purification of recombinant proteins. SDS gels stained with Coomassie Blue showing the indicated purified recombinant proteins used for protein-protein assays. The GST-hTAFII1351-372 is a mutant lacking the 372 amino acids at the N-terminal of the protein and containing the HFM.

View larger version (76K): [in this window] [in a new window]   Fig. 4.   Protein-protein interactions of TAFIIs and NF-Y HFM subunits. A, in the different panels, wt NF-YB-NF-YC His-tagged were incubated with GST-hTAFII18 and GST-hTAFII28, before loading onto NTA-agarose. Load, flow-through, 300 mM, and 1.0 M KCl washes and the two imidazole-eluted fractions are indicated. The fractions were run in SDS gels, and the proteins are revealed by Western blotting with the indicated antibody. Similar experiments are shown for GST-hTAFII20, GST-hTAFII1351-372, GST-hTAFII20-GST-hTAFII1351-372, and hTAFII80-hTAFII31. In the lower panels, an NF-YB-NF-YC dimer devoid of the His tag, consisting of wt NF-YB and of the HFM-containing YC5 mutant (54), was incubated with the indicated hTAFIIs. From top to bottom: His-tagged hTAFII28, His-tagged hTAFII18, and hTAFII28-hTAFII18 (both His-tagged).

From this set of experiments we conclude that only hTAF_II28, hTAF_II18-hTAF_II28, and hTAF_II20, but not the HFM containing hTAF_II20-hTAF_II1351-372, nor hTAF_II80-hTAF_II31, are capable to associate the NF-Y HFM dimer in solution.

Binding of NF-Y to Histone Fold TAF_IIs-- Because NF-YA is necessary for CCAAT-box binding and is known to recognize determinants in the HFMs of both NF-YB and NF-YC (19), it was important to test whether the interactions with TAF_IIs would also be scored in the context of the trimeric complex. NF-YA has an intrinsic affinity for NTA-agarose and is unsuitable for the protein-protein interaction approach taken above, most likely because of the high number of His residues in the conserved domain (Not shown). We thus switched to immunoprecipitations with the purified recombinant NF-Y trimer and the different TAF_IIs, either as single subunits or dimers. The complexes were incubated with Mab7, a monoclonal antibody that recognizes the NF-YA Q-rich activation domain (55), previously bound to a Protein G-Sepharose matrix; in parallel, equivalent amounts of recombinant proteins were incubated with a control Protein G-Sepharose resin associated with an irrelevant anti-MHC class II antibody. After washing, bound material was recovered by boiling samples in SDS buffer and analyzed in Western blots with the anti-NF-Y and anti-TAF_II antibodies. As expected, NF-YA and NF-YB are immunoprecipitated with Mab7 but not with the control antibody (Fig. 5A, upper panel); the same was true for NF-YC (not shown, see below). Incubation of NF-Y with hTAF_II28, but not with TAF_II18, retained the TAF_II in the bound material. Surprisingly, unlike the previous experiments with the NF-YB-NF-YC dimer, when incubated together with the NF-Y trimer, interactions of the hTAF_II28-hTAF_II18 dimer were not observed, implying that the presence of NF-YA prevents hTAF_II28-hTAF_II18 binding to the NF-Y HFM dimer. On the other hand, hTAF_II20 and the hTAF_II80-hTAF_II31 dimer, but not hTAF_II20-hTAF_II1351-372, were bound to the NF-Y trimer. hTAF_II80, but not hTAF_II31, was immunoprecipitated when incubated alone with NF-Y. In this experimental setting, we also used a differentially spliced form of hTAF_II80, termed hTAF_II80, in which 10 amino acids of the HFM 2 are missing. This isoform is incapable of interacting with hTAF_II31.² We tested this protein in the immunoprecipitation assays and found that, as for hTAF_II80, it is able to interact with NF-Y. To confirm that hTAF_II20-hTAF_II1351-372 and hTAF_II28-hTAF_II18 are present in dimeric form in our assays, we immunoprecipitated these dimers, in the absence of NF-Y subunits, with anti-hTAF_II135 and anti-hTAF_II18 monoclonal antibodies, respectively. Fig. 5B shows that both hTAF_II20 and hTAF_II28 are indeed associated with their respective partners, as assessed in Western blots.

View larger version (51K): [in this window] [in a new window]   Fig. 5.   Immunoprecipitations of NF-Y and HFM TAFIIs. A, equivalent amounts of recombinant NF-Y and the indicated hTAFIIs were incubated and immunoprecipitated with the anti-NF-YA monoclonal antibody 7 (55), or with an anti-MHC class II control antibody. Load (L), unbound (U), and bound (B) materials were loaded onto SDS gels and checked in Western blots with the indicated antibody. B, Western blots of immunoprecipitations of the His-hTAFII18-His-hTAFII28 dimer with the anti-hTAFII18 antibody 16TA. Similarly, Western blot analysis of immunoprecipitations of preparations containing GST-hTAFII20 and GST-hTAFII1351-372 with the 20TA monoclonal antibody against hTAFII135.

From the immunoprecipitation analysis of the NF-Y trimer, we conclude that hTAF_II18 and hTAF_II31 do not bind NF-Y, whereas hTAF_II80, hTAF_II80, hTAF_II28, and hTAF_II20 are capable of associating with the trimer. Upon dimerization, hTAF_II31-hTAF_II80, but not hTAF_II28-hTAF_II18, can bind to NF-Y.

Effect of TAF_IIs Overexpression on NF-Y Activation in Mammalian Cells-- Having established that NF-Y is capable of increasing the binding of holo-TFIID and associated multiple TAF_IIs in vitro, we sought to investigate the in vivo effects with "co-activator" assays used for other activators (12, 59-62). The system is based on the co-expression in NIH-3T3 fibroblasts of GAL4-NF-YA together with NF-YB and NF-YC, activating a promoter containing five GAL4 sites driving the Luciferase reporter gene: Transcription is strictly dependent upon co-transfection of all NF-Y subunits, requiring at least one of the NF-Y Q-rich domains (56; see Fig. 6A). Co-transfections of different amounts of vectors expressing the TAF_IIs used in our in vitro analysis gave the results outlined in Fig. 6B: hTAF_II20 and hTAF_II135 had small, 2-fold-positive effects on GAL4-NF-Y activation; hTAF_II31 and hTAF_II80, either alone or in combination showed no effect; whereas expression of hTAF_II18 and hTAF_II28, alone or together, had a clear dose-dependent negative effect. To verify whether this inhibition was specific for the NF-Y trimer, or exerted indirectly through the basal promoter, we tested in similar assays an NF-YA mutant, G4-YA12, containing the isolated activation domain of NF-YA fused to the DNA-binding domain of GAL4, and GAL4-SP1, also a Q-rich activator. Co-expression of hTAF_II18 and/or hTAF_II28, or of the other TAF_IIs, had minor effects on GAL4-YA12 or GAL4-SP1 (Fig. 6, C and D): if anything, hTAF_II18 and/or hTAF_II28 slightly increased GAL4-SP1 activation. Because the reporter construct was identical in all these experiments, these latter experiments rule out that the strong inhibitory effect of hTAF_II18-hTAF_II28 observed on the GAL4-NF-Y trimer is due to an unspecific repression of core promoter activity; moreover, inhibition is specific for the NF-Y trimer and not for the Q-rich activation domains, such as those of NF-YA, or of SP1.

View larger version (33K): [in this window] [in a new window]   Fig. 6.   TAFIIs-mediated transcriptional modulation of NF-Y activity. A, scheme of the GAL4 vectors used in the transfection experiments (56). Black boxes indicate the GAL4 DNA-binding domain; hatched boxes represent the Q-rich regions; gray boxes the homology domains. B, NIH-3T3 fibroblasts were transfected with a GAL4-driven Luciferase reporter and 1 or 3 µg of TAFIIs plasmids as indicated. C and D, same as B, except that GAL-SP1 and GAL4-YA12 (56) were used with the indicated co-transfected TAFIIs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

TFIID Recruitment-- The TATA-box is the most frequent promoter element. When absent, polymerase II positioning is in general assured by an initiator; i.e. both these core elements are recognized by TFIID (1). NF-Y binding sites are found in 25% of eukaryotic promoters, invariably playing an important and sometimes essential role. Both boxes are normally found at a fixed position, TATA at 25/30 and CCAAT at 60/100. The mean CCAAT-box position and orientation in TATA and TATA-less promoters pointed to a small, but significant difference (20),³ suggesting that the proteins binding to these sites could directly or indirectly interplay. In keeping with this, we previously presented biochemical evidence showing that the NF-Y HFM subunits are associated with holo-TFIID and can bind TBP directly (41). The aim of our study was to examine their relationships, using a model system, the MHC class II Ea promoter, in which the functional importance of the two complexes is well established. We have previously shown, in fact, that holo-TFIID binds to a TdT-like initiator in the Ea promoter in a sequence-specific way. Our EMSA analysis is the first indication that NF-Y helps recruit holo-TFIID, thereby adding experimental proof to the hypothesis that NF-Y and TFIID have intimate relationships. It should be noted that other upstream factors are thought to recruit TBP and associated factors, as an important step in the formation of a transcriptional competent complex (1). Concerning the mechanisms of such DNA-binding facilitation, an obvious hypothesis is that there are direct protein-protein contacts between TAF_IIs and NF-Y. We already detailed the binding of NF-YB-NF-YC to TBP (41), and the in vitro analysis presented here is strongly supportive of this, because a number of TAF_IIs show affinity for NF-Y. Indeed, the multiple interactions of HFM TAF_IIs with NF-Y observed in solution invite the speculation that TFIID binding might be influenced, if not dictated, by one or more transcription factor combinations binding nearby. Interestingly, we evidenced holo-TFIID complexes showing differential mobilities in our EMSAs, depending on the presence or absence of NF-Y: The upper TFIIDb, for example, migrates faster with NF-Y, possibly suggesting that a different composition of subunits might be present and that a DNA-bound NF-Y might select specific sub-complexes. The apparent multiplicity of TFIID combinations could then reflect into a plasticity of DNA binding, because the presence of adjacent activators might select subtypes of holo-TFIIDs with subunit compositions particularly suited to fit within the context of a given promoter. Antibody supershifting experiments, although showing that "core" TAF_IIs are present in the IID/Y complexes, cannot yet provide us with a complete description of the composition of these complexes.

TFIID is known to be capable of sequence-specific interactions, and multiple subunits within TFIID can contact DNA (63, 64 and references therein). In the course of our analysis we found no evidence of sequence-specific binding of the isolated HFM TAF_IIs considered here to the Ea promoter (data not shown). Non-HFM TAF_IIs with well documented core promoter DNA-binding specificity, such as hTAF_II150 and/or hTAF_II250 (64), could then be considered for the Inr-binding activity.

NF-Y Interactions with HFM TAF_IIs-- Complete reconstitution of the holo-TFIID complex with recombinant subunits has not been achieved yet. We therefore dissected NF-Y interactions with individual TAF_IIs. Given the common structural features of many TAF_IIs with NF-Y subunits, we decided to start with HFM TAF_IIs. In general, our analysis showed several interactions between NF-Y and TAF_IIs: Minimally, hTAF_II28, hTAF_II80, and hTAF_II20 have affinity for either the HFM NF-YB-NF-YC dimer, the trimer, or both (Table I). hTAF_II135 should be added to this list (42), because the negative results obtained here probably reflect the absence of the N-terminal Q-rich region. Therefore, the interactions of the trimer with hTAF_II20 and hTAF_II80 are possibly relevant for the function of the CCAAT-binding trimer. On the other hand, the interactions of hTAF_II28-hTAF_II18 with the HFM NF-YB-NF-YC dimer, but not with the trimer, should be considered in the light of the presence of the HFM dimer, without NF-YA, in several cell types, including monocytes and differentiated myotubes (21). This finding might point to a role in the basic mechanisms of activation on promoters that lack the CCAAT target site.

The HFM TAF_IIs are present not only in TFIID but also in other complexes: (i) hTAF_II135, hTAF_II80, and hTAF_II20 are found in TFTC, a TAF_IIs-containing complex lacking TBP (27, 30); (ii) hTAF_II20 and hTAF_II31 (and the hTAF_II80-like PAF65) are in the P/CAF complex (18); and (iii) hTAF_II31 is found in STAGA (28). A histone octamer-like structure within TFIID has been hypothesized (10); indeed, core histones and hTAF_II80, hTAF_II31, and hTAF_II20 interact through their HFMs in ways that are consistent with histones rules: The H4-like hTAF_II80 with H3 and H2B, the H3-like hTAF_II31 with H4, and the H2B-like hTAF_II20 with H2A and H4. These interactions are fully in agreement with previous findings on the binding of H3-H4 and H2A-H2B subfamilies of histone folds. Crystallographic analysis of the nucleosome found details for residues that are required for H4-H2B interactions: H4 His-75 and H4 Lys-91, and H2B Glu-90 and H2B Glu-73 (5). These residues are conserved in hTAF_II80 and in the related PAF65. In NF-YB, the Asp-115 and Glu-98 residues corresponding to H2B Glu-90 and Glu-73 are among the relatively few amino acids present in all 26 NF-YB sequences from different species (65). We found that NF-Y is able to interact with an isoform of hTAF_II80-hTAF_II80, which lacks 10 amino acids in the 2 helix of the HFM (66), a fact that is mirrored by the interaction of hTAF_II80 with the H2B-like hTAF_II20.⁴ Under these circumstances, it is unclear what the structure of the HFM might be, but the 2 subunit is crucial for the formation of heterodimers and, indeed, hTAF_II80 is unable to bind to hTAF_II31: Other parts of the HFM might be involved in the contacts with the H2B-likes. Alternatively, a domain of hTAF_II80 distinct from the HFM might be implicated in contacting NF-Y. Concerning the hTAF_II28/hTAF_II18 dimer, structural analysis detailed somewhat different sorts of HMFs not easily assigned to any of the core histone sub-classes (15). Thus it is difficult to rationalize our interactions data with available structural information.

TAF_IIs are known to contact the activation domains of gene-specific upstream factors, and indeed the capacity of a given factor to activate transcription in vitro correlates well with its TAF_II-binding ability (1). Conditional inactivation of yeast histone fold TAF_IIs, yTAF_II17/hTAF_II31, yTAF_II60/hTAF_II80, and yTAF_II68/hTAF_II20, provided compelling genetic evidence for their general role in promoter activation (34-38). However, only a limited set of studies focused on the effect of TAF_IIs overexpression in mammalian cells. In some cases, positive effects were seen: hTAF_II28 on the activation factor 2 of retinoic X receptor and hTAF_II135 on retinoic acid receptor, vitamin D receptor and thyroid receptor (12, 59). In other reports, in vitro interactions, SP1-hTAF_II135, E1A-hTAF_II135, and p53-dTAF_II40-dTAF_II60-dTAF_II230, were matched by strong repression in co-transfections of the TAF_IIs with GAL4 fusions containing the activation domains (61, 67), although the same TAF_IIs interactions with GAL4-p53 resulted in activation in vitro (68). In our GAL4 assays, we observe a similar negative effect by overexpressing hTAF_II28 and/or hTAF_II18 on the activation of the NF-Y trimer. It should be noted that this is specific, both for the target, it is not seen with GAL4-SP1 and the Q-rich activation domain of NF-YA, and for these two TAF_IIs, because hTAF_II31-hTAF_II80 have negligible effects and hTAF_II20-hTAF_II135 have small positive effects; moreover, hTAF_II28-hTAF_II18 do not inhibit the natural Ea promoter (not shown). What might be the reason for the inhibition of GAL4-NF-Y fusions? Our in vitro results indicate that the hTAF_II28-hTAF_II18 dimer interacts with NF-YB-NF-YC, but not with the NF-Y trimer, implying that NF-YA, which recognizes determinants in the HFMs of both subunits (20), prevents the association of hTAF_II28-hTAF_II18 to NF-YB-NF-YC. Because formation of the trimer in vivo is essential for GAL4-NF-YA activation (56), one can imagine that the co-transfected TAF_IIs could compete for binding to NF-YA by associating NF-YB-NF-YC. However, other interpretations must reconcile our findings that the NF-Y-interacting hTAF_II28 and the non-interacting hTAF_II18 both inhibit when transfected alone. It is possible that overexpression of some TAF_IIs alters the stoichiometry of endogenous TFIID complexes, impairing their capacity to mediate activation through the artificial GAL4 constructs. In this respect, it should be noted that certain mouse tissues do have with lower amounts of hTAF_II28 and hTAF_II18 (33).

TAF_IIs and MHC Class II Transcription-- In addition to NF-Y, binding the trimeric complex RFX is important for MHC class II promoters (53). The NF-Y-RFX trimers make cooperative interactions. In particular, NF-Y binding improves the otherwise rather inefficient binding of RFX. Two non-DNA-binding co-activators are also crucial: the ubiquitous p300/CBP (69, 70) and the tissue-specific CIITA (reviewed in Ref. 71). A network of protein-protein interactions is emerging, in particular, CIITA can interact with hTAF_II31. Our finding that the hTAF_II80-hTAF_II31 dimer can associate NF-Y through hTAF_II80 suggests that hTAF_II80-hTAF_II31 could be contacted at the same time by two of the MHC class II activators. The interactions of NF-Y with HFM TAF_IIs should be considered in the light of recent findings showing that both complexes can associate HATs: p300, P/CAF, and hGCN5 interact with NF-Y (21), whose NF-YB subunit is acetylated by p300 (72). Similarly, CBP/p300 can interact with CIITA. It has been suggested that TAF_IIs could be chaperones of the histone-modifying machines in the proximity of core promoters. NF-Y could recruit acetylase complexes not only through direct interactions with P/CAF, GCN5, or CBP/p300, but also by contacting P/CAF complexes, via the HFM TAF_IIs. From its privileged location at 60, NF-Y is clearly a pivotal factor at the cross-road of multiple connections: It helps upstream factors such as RFX on MHC class II promoters bind DNA, and it recruits TFIID by binding to TBP and to core TAF_IIs. At a higher level, NF-Y reaches its site efficiently in the context of a pre-formed nucleosome, interfacing well with H3-H4 tetramers (54). In summary, NF-Y represents an excellent candidate for penetrating chromatin structures, allowing other upstream activators bind their sites, organizing TFIID complexes and recruiting co-activators that further modify, by acetylation, surrounding nucleosomes. Studies aimed at clarifying the complexity of NF-Y-TFIID interplay by taking into account the interactions with other TAF_IIs are currently underway.

	ACKNOWLEDGEMENTS

We are grateful to A. Levine, L. Lania, and G. Caretti for gift of reagents. We thank M. Minuzzo for technical assistance.

	FOOTNOTES

^* This work was in part supported by grants from Ministero dell'Università della Ricerca Scientifica e Tecnologica (PRIN 99) and Associazione Italiana per la Ricerca sul Cancro (to R. M.).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.

^¶ To whom correspondence should be addressed: Tel.: 39-059-205-5542; Fax: 39-059-205-5548; E-mail: mantor@mail.unimo.it.

Published, JBC Papers in Press, October 31, 2001, DOI 10.1074/jbc.M103651200

² B. Bell and L. Tora, submitted.

³ R. Mantovani, unpublished.

⁴ B. Bell and L. Tora, unpublished.

	ABBREVIATIONS

The abbreviations used are: HFM, histone fold motif; NF-Y, nuclear factor Y; TAF_IIs, TBP-associated factors PO1II; wt, wild type; EMSA, electrophoretic mobility shift assay; NTA, nitrilotriacetic acid; GST, glutathione S-transferase; MHC, major histocompatibility complex; TBP, TATA binding protein; P/CAF, P300 CBP associated factor; SP1, SP protein 1; RFX, DR factor X; Inr, initiator; STAGA, SPT3-TAFII31-GCN5-L acetyltransferase; TFTC, TBP-free TAF complex; CIITA, class II transcriptional activator.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This article has been cited by other articles: