The TFIID Components Human TAF_II140 and Drosophila BIP2 (TAF_II155) Are Novel Metazoan Homologues of Yeast TAF_II47 Containing a Histone Fold and a PHD Finger

Abstract

The RNA polymerase II transcription factor TFIID comprises the TATA binding protein (TBP) and a set of TBP-associated factors (TAF_IIs). TFIID has been extensively characterized for yeast, Drosophila, and humans, demonstrating a high degree of conservation of both the amino acid sequences of the constituent TAF_IIs and overall molecular organization. In recent years, it has been assumed that all the metazoan TAF_IIs have been identified, yet no metazoan homologues of yeast TAF_II47 (yTAF_II47) and yTAF_II65 are known. Both of these yTAF_IIs contain a histone fold domain (HFD) which selectively heterodimerizes with that of yTAF_II25. We have cloned a novel mouse protein, TAF_II140, containing an HFD and a plant homeodomain (PHD) finger, which we demonstrated by immunoprecipitation to be a mammalian TFIID component. TAF_II140 shows extensive sequence similarity to Drosophila BIP2 (dBIP2) (dTAF_II155), which we also show to be a component of Drosophila TFIID. These proteins are metazoan homologues of yTAF_II47 as their HFDs selectively heterodimerize with dTAF_II24 and human TAF_II30, metazoan homologues of yTAF_II25. We further show that yTAF_II65 shares two domains with the Drosophila Prodos protein, a recently described potential dTAF_II. These conserved domains are critical for yTAF_II65 function in vivo. Our results therefore identify metazoan homologues of yTAF_II47 and yTAF_II65.

Transcription factor TFIID is one of the general factors required for accurate and regulated initiation by RNA polymerase II. TFIID comprises the TATA binding protein (TBP) and TBP-associated factors (TAF_IIs) (3, 24). A subset of TAF_IIs are present not only in TFIID but also in the SAGA, PCAF, STAGA, and TFTC complexes (5, 16, 22, 33, 38, 53).

The function of TAF_IIs has been studied using in vitro transcription systems which have provided evidence that they act as specific coactivators by interacting directly with transcriptional activator proteins (48). Such studies also indicated that TAF_IIs contribute to promoter recognition and promoter selectivity (8, 10, 49). Genetic evidence obtained from TAF_II knockout and depletion experiments and temperature-sensitive (TS) mutants indicates that TAF_IIs are involved in the control of gene expression in yeast and mammalian cells and that they act as antiapoptotic factors (11, 23, 37; for reviews, see references 2, 3, and 23).

TFIID has been characterized for Drosophila, humans, and yeast (7, 14, 40, 56). After its initial purification, many subunits were rapidly identified and the corresponding genes were cloned. Sequence comparisons allowed the identification of TAF_IIs in each species sharing one or more conserved domains. Many of these domains correspond to structural motifs, such as the histone fold domain (HFD), the β-transducin repeat, and bromodomain, or catalytic domains, such as the histone acetyltransferase domain. These results indicate that the composition and organization of TFIID has been well conserved between yeast and humans (reviewed in reference 16).

In recent years, it has been assumed that all the mammalian TFIID components have been identified. Several observations, however, challenge this assumption. Characterization of highly purified yeast TFIID (yTFIID) has identified two novel TFIID subunits, yTAF_II48 (Mpt1) and yTAF_II65 (42, 45). While yTAF_II48 is the homologue of human TAF_II135 (hTAF_II135) in yeast (18, 42), no obvious metazoan homologues of yTAF_II65 have yet been identified. TAF_II65 contains an HFD which selectively heterodimerizes with yTAF_II25 (17). Furthermore, a second yTAF_II, yTAF_II47 (52), contains an HFD and selectively heterodimerizes also with yTAF_II25 (17). At present, no homologues of yTAF_II47 and yTAF_II65 in metazoans have been identified, yet a yTAF_II25 homologue has been well characterized for humans (hTAF_II30 [27]), and for Drosophila, two homologues have been identified (Drosophila TAF_II24 [dTAF_II24] and dTAF_II16 [19]). Metazoan TFIID must therefore contain a heterodimerization partner(s) for hTAF_II30, and dTAF_II16, and dTAF_II24, which would be considered homologues of yTAF_II47 and/or yTAF_II65.

In database searches with the yTAF_II47 HFD, we previously identified metazoan proteins with HFDs similar to that of yTAF_II47 (17). Here, we describe mouse TAF_II140 (mTAF_II140) and hTAF_II140, novel proteins containing an HFD with strong similarity to that of yTAF_II47 and a plant homeodomain (PHD) finger. We show that the mouse and human TAF_II140 HFDs selectively and directly heterodimerize with that of hTAF_II30. Immunoprecipitation experiments confirm that hTAF_II140 is a bone fide TAF_II which can be immunopurified from HeLa cell extracts under stringent conditions with TBP and other identified hTAF_IIs. Mammalian TAF_II140 shows high sequence similarity to the Drosophila protein BIP2 (bric-a-brac interacting protein 2), which also contains an HFD and PHD finger. We show that the dBIP2 HFD selectively and directly heterodimerizes with dTAF_II24 and that BIP2 is a component of dTFIID.

The Drosophila Prodos (PDS) is a protein essential for cell viability which has been shown to comprise an HFD which selectively heterodimerizes with dTAF_II16. Consequently, it has been proposed that PDS may be a dTFIID component (25). Here, we show that PDS shares sequence similarity with yTAF_II65 both in the HFD and in a second domain (designated the TAF_II65-PDS [TAP] domain [TAPD]) C terminal of the HFD. We show that the yTAF_II65 HFD and the TAPD are partially redundant functional domains required for yeast viability. The fact that PDS shares the HF and TAP domains which are required for yTAF_II65 function in vivo, along with its observed heterodimerization with dTAF_II16, indicates that PDS is the Drosophila homologue of yTAF_II65.

MATERIALS AND METHODS

Cloning of mTAF_II140 and hTAF_II140.

The cDNA encoding mTAF_II140 HFD was isolated by reverse transcription (RT)-PCR using mouse F9 embryonal carcinoma cell RNA and oligonucleotide primers based on expressed sequence tag (EST) sequences. To isolate the full-length cDNA, an F9 embryonal carcinoma cDNA library was screened with the ³²P-labeled HFD fragment. A clone comprising the 5′ untranslated region (UTR) and amino acids 1 to 340 was isolated. A second overlapping clone encoding amino acids 238 to 747 was isolated by screening the same library with oligonucleotide probes derived from EST sequences. The 3′ end of the cDNA was cloned by RT-PCR using an oligonucleotide primer at amino acid 651 and primers in the 3′ UTR derived from EST sequences. The partial cDNA sequences were cloned in pBSK and sequenced. The three cDNAs were assembled by overlapping PCR into two partially overlapping fragments. In one fragment, a BamHI site was introduced at nucleotide 1286 (GGCTCA-GGATCC) without changing the amino acid sequence. This fragment was cloned into the EcoRI-BamHI sites of expression vector pXJ41 (54). A fragment from nucleotide 1286 to the end of the coding sequence was amplified with primers containing BamHI and XhoI sites, and the resulting fragment was cloned into the pXJ41 vector containing the 5′ fragment, thus reconstituting the full open reading frame. The hTAF_II140 cDNA fragment containing the 5′ UTR and genes for amino acids 1 to 727 was cloned by screening a HeLa cell cDNA library with an oligonucleotide corresponding to genes for amino acids 54 to 72 of the HFD based on EST sequences. All constructs were verified by automated DNA sequencing, and further details are available upon request.

Preparation of antibodies and immunoprecipitations.

Antibodies against human and mouse TAF_II140 were generated against a synthetic peptide corresponding to amino acids 108 to 127 of hTAF_II140 (as indicated in Fig. 2) coupled to ovalbumin. Monoclonal antibodies were generated as previously described (32). Two resulting monoclonal antibodies, 1C7 and 2F5, were used. The anti-BIP2 polyclonal antiserum was collected from rabbits immunized with a bacterially expressed glutathione S-transferase (GST)-fusion protein comprising amino acid residues 719 to 909 of the BIP2 protein. The anti-SAP130 and GCN5 antibodies were as previously described (M. Brand, J. G. Moogs, F. Lejeune, J. Stevenin, M. Oulad-Abdelghani, G. Almouzni, and L. Tora, submitted for publication). The anti-TBP monoclonal antibodies 2C1 and 3G3 and the anti-TAF_II monoclonal antibodies were as previously described (6, 13, 27, 32, 35, 36). The polyclonal antisera against dTBP, dTAF_II16, dTAF_II24, and dTAF_II230 were as previously described (19, 28). Immunoprecipitations of HeLa cell nuclear extracts with monoclonal antibody 2C1 and elutions with the epitope peptide were performed as previously described (6, 27, 35, 36). Immunoprecipitations of Drosophila embryo extracts were performed as previously described (19). Western blottings were performed by standard techniques, and the proteins were detected using an ECL kit (Amersham) and autoradiography.

FIG. 2.

The amino acid sequences of mTAF_II140 and hTAF_II140 are aligned. The positions of the HFD and PHD fingers are shown. In the PHD finger, the critical cysteine and histidine residues are shown in white letters. The amino acid sequences used to generate monoclonal antibodies are underlined. The alignment was generated using the GCG program.

Two-hybrid and coexpression assays.

All yeast and bacterial expression vectors were constructed by PCR using primers with the appropriate restriction sites, and constructs were verified by automated DNA sequencing. Details of constructions are available on request. LexA fusions were constructed in the multicopy vector pBTM116 containing the TRP1 marker and the VP16 fusions were constructed in the multicopy vector pASV3 containing the LEU2 marker (18), and assays were performed with the L40 strain as previously described (17, 50). Quantitative β-galactosidase assays on individual L40 transformants were determined as previously described (18). Reproducible results were obtained in several independent experiments, and the results of a typical experiment are shown in the figures.

Yeast complementation assays.

Complementation assays were performed by the plasmid shuffle technique. Wild-type or mutated yTAF_IIs were cloned in the multicopy pAS3 plasmid with a LEU marker. All yeast strains were transformed by the LiAc technique. Yeast strain YSLS58 that was used for plasmid shuffling of TAF65 was derived from YSLS41 (45) by sporulation and tetrad dissection as previously described (17). For complementation assays, the indicated rescue plasmids were transformed and the wild-type TAF (URA3) plasmid was shuffled out by two passages on media containing 5-fluoroorotic acid. In all experiments, cultures were grown in yeast-peptone-dextrose unless selection was necessary, in which case cultures were grown in the appropriate selective synthetic dextrose (SD) medium.

Immunostaining of Drosophila embryos.

The 0- to 2-h and 2- to 24-h embryos were dechorionated, and antibody staining and diaminobenzidine visualization were performed as previously described (44). The anti-BIP2 polyclonal antibody was diluted 1/3,000, and the peroxidase-conjugated goat anti-rabbit immunoglobulin G (IgG) was diluted 1/1,000.

Immunostaining of Drosophila salivary gland polytene chromosomes.

Squashes of polytene chromosomes and antibody stainings were performed as previously described (57). The anti-BIP2 primary antibody was diluted 1/200. The Cy3-conjugated secondary antibodies were diluted 1/500. DNA was stained with Hoechst 33258.

Nucleotide sequence accession numbers.

The sequences reported here have been assigned the following EMBL database accession numbers: mTAF_II140, AJ292189; hTAF_II140, AJ292190; dTAF_II155, AJ292191.

RESULTS

HFD of mTAF_II140 heterodimerizes with hTAF_II30.

We previously reported the existence of a putative HFD at the N terminus of yTAF_II47 which is necessary and sufficient for yeast viability (17). This HFD shows strong sequence similarity to those of the hTAF_II135/ADA1/H2A family and mediates a selective heterodimerization with the yTAF_II25 HFD (17). Database searches with the yTAF_II47 HFD identified several metazoan proteins with potential HFDs located at their N termini (reference 17 and Fig. 1A).

FIG. 1.

Functional analysis of the HFD from a putative mouse and human yTAF_II47 homologue. (A) Alignment of the HFD sequence of yTAF_II47 with those of its putative Candida albicans (Can), zebrafish (Danio rerio), Drosophila melanogaster (dBIP2), and human and mouse homologues. The positions of the predicted α-helices and loops are indicated above the sequences. Identical amino acids are shown in white letters on a black background. Positions with conserved, mainly hydrophobic amino acids are boxed in gray. Amino acids were classified as follows: small residues, P, A, G, S, T; hydrophobic residues, L, I, V, A, F, M, C, Y, W; polar and acidic residues, D, E, Q, N; basic residues, R, K, H. Threonine residues are occasionally present in otherwise hydrophobic positions. The amino acid sequences shown without numbers are predicted from genomic (gen) or EST sequences. The accession numbers for the indicated sequences are as follows: C. albicans, 396380B03; zebrafish, GenBank accession no. AW343321fi76b06.y1; Drosophila BIP2, Q9XZU7; mouse, GenBank accession no. AA692266ur52c07. (B) Graphical representation of quantitative two-hybrid β-galactosidase assays. The LexA chimeras shown above the graph were tested with the VP16 chimeras shown below each column. −, negative controls with the VP16 domain alone. (C, D, and E) Coexpression of the yTAF_II47 and mTAF_II140 HFDs with those of other TAF_IIs in E. coli cells. Bacteria were transformed to express the proteins shown above each lane, and following extract preparation, the soluble protein retained on glutathione-Sepharose beads was analyzed by SDS-PAGE and staining with Coomassie brilliant blue. /, GST-fusion was expressed alone. The locations of the GST-yTAF_II and GST-mTAF_II fusions and the retained yTAF_II25 and hTAF_II30 proteins are indicated.

To determine whether the HFD, which is identical in the human and murine proteins, would heterodimerize with hTAF_II30, two-hybrid experiments were performed in yeast by using LexA fusions of the hTAF_II30 HFD and VP16 fusions of the mouse protein (hereafter designated mTAF_II140). As previously described (17), strong interactions between the HFDs of yTAF_II47 and yTAF_II25 or hTAF_II30 were observed (Fig. 1B, lanes 1 and 5). Similarly, the mTAF_II140 HFD also interacted strongly with both the yTAF_II25 and hTAF_II30 HFDs (Fig. 1B, lanes 2 and 6). These interactions were abolished by mutation of residue W23 (W23R) in the L1 loop of mTAF_II140 HFD [mTAF_II140(1-88)m1] (Fig. 1B, lanes 3 and 7).

Direct heterodimerization of the mTAF_II140 and hTAF_II30 HFDs was tested by coexpression in E. coli (17, 18). When expressed alone, a GST fusion of the yTAF_II47 or mTAF_II140 HFDs was poorly soluble [Fig. 1C and D, lane 1, GST-yTAF_II47(1-81) and GST-mTAF_II140(1-88)]. In contrast, when coexpressed with the yTAF_II25 HFD, GST-yTAF_II47(1-88) was solubilized and formation of a heterodimeric complex was observed (Fig. 1C, lane 2, and reference 17). Similarly, GST-mTAF_II140(1-88) was also solubilized when coexpressed with the yTAF_II25 and hTAF_II30 HFDs, and the formation of a heterodimeric complex was observed in each case (Fig. 1D, lanes 2 and 3, and Fig. 1E, lane 2). Solubilization by the hTAF_II30 HFD and complex formation was abolished by the W23R mutation, which eliminated interaction in yeast (Fig. 1E, lane 4). No heterodimerization was seen, however, when the yTAF_II47 or mTAF_II140 HFDs were coexpressed with the HFDs of other TAF_IIs (Fig. 1C, lanes 3 to 5, Fig. 1D, lanes 4 to 6, and data not shown).

Together, these results indicate that the HFD of mTAF_II140 selectively and directly heterodimerizes with hTAF_II30, indicating that it may be a novel component of mammalian TFIID.

TAF_II140 is a HeLa cell TFIID subunit.

The full-length mTAF_II140 open reading frame was cloned by a combination of screening an F9 cell cDNA library and RT-PCR using information from overlapping clones in the EST databases (see Materials and Methods). Mouse TAF_II140 comprises 932 amino acids with the HFD located between amino acids 9 and 72 (Fig. 2). Interestingly, mTAF_II140 also possesses a single PHD finger of the type Cys4-His-Cys3 originally described for the HAT3.1, Polycomb-like, and HRX proteins (1, 47) (sometimes also called the leukemia-associated protein [LAP] domain [43]) at the C terminus of the protein. This PHD finger is preceded by a highly charged region comprising multiple lysine and glutamic acid residues and a region rich in proline and other small residues (Fig. 2).

A partial cDNA encoding the first 727 amino acids of hTAF_II140 was also isolated by screening a HeLa cell cDNA library. Comparison with mTAF_II140 showed that the HFD was identical in the two proteins which otherwise share strong sequence similarity (Fig. 2). Although we could not isolate a full-length clone for hTAF_II140, database searches identified a genomic clone and ESTs encoding a highly related PHD finger followed by a stop codon, as in mTAF_II140 (Fig. 4B). This provides strong evidence for the existence of an hTAF_II140 protein with an overall structure analogous to that of mTAF_II140.

FIG. 4.

(A) Alignment of the mTAF_II140 and dBIP2 amino acid sequences. The locations of the histone fold and PHD fingers are indicated. (B) Comparison of the PHD fingers of mTAF_II140, hTAF_II140, and dBIP2 with some of those previously characterized. Identical amino acids are shown in white letters against a black background. Critical cysteine and histidine residues are indicated (∗).

Monoclonal antibodies were raised against a peptide of hTAF_II140 downstream of the HFD (Fig. 2, underlined), which is almost perfectly conserved between humans and mice. This monoclonal antibody recognizes a protein with a molecular mass of around 140 kDa in Cos cells transfected with an mTAF_II140 expression vector (Fig. 3A, lane 2). Although no endogenous protein was seen with small amounts of control untransfected Cos cell extract (Fig. 3A, lane 1), hTAF_II140 was clearly detected in more-concentrated HeLa cell nuclear extract (lane 3). The antibody also recognized a nuclear protein in HeLa cells in immunofluorescence experiments (data not shown).

FIG. 3.

Immunoprecipitations. (A) The origin of each extract is shown above each lane. Cos Untrans, 1 μg of untransfected Cos cell extract; Cos Trans, 1 μg of extract from Cos cells transfected with 5 μg of pXJ41-mTAF_II140; HeLa N.E., 15 μg of HeLa cell nuclear extract; IP anti-TBP, TFIID immunoprecipitated with monoclonal antibody 2C1 and eluted with the corresponding peptide. The positions of mTAF_II140 and hTAF_II140 are indicated along with that of residual 2C1 IgGH detected by the secondary conjugated antibody used in ECL. All lanes were probed with anti-TAF_II140 antibody 1C7. (B) The layout is the same as that in panel A, with the origin of the protein indicated above each lane. E1 and E2 indicate the first and second elutions of the 2C1 immunoprecipitation with the epitope peptide. The first lane shows a control immunoprecipitation with anti-Flag antibody m2. Lanes 1 to 3 were probed with the anti-TBP antibody 3G3 and the anti-hTAF_II140 antibody 2F5. Lanes 4 and 5 show the same blots that are in lanes 2 and 3, probed with additional antibodies; lane 6 shows HeLa cell nuclear extract incubated with all of the antibodies. (C) HeLa cell nuclear extract was precipitated with monoclonal antibody 1C7, and the precipitates eluted with the epitope peptide. The blot was probed with the antibodies used to detect the proteins indicated. In lanes 3 and 4, the blot shown in lanes 1 and 2 was reprobed with the indicated antibodies after inactivation of the first signal by prolonged incubation with ECL.

To determine whether hTAF_II140 was a component of immunopurified TFIID, HeLa cell extracts were precipitated with the anti-TBP monoclonal antibody 2C1 (32) and the TFIID was eluted by using the 2C1 epitope peptide. HeLa cell TAF_II140 was clearly detected in the eluted TFIID (Fig. 3A, lane 4). HeLa cell TAF_II140 was also detected in immunopurified TFIID by using a second independent monoclonal antibody against the same peptide (Fig. 3B, lanes 2 and 3), while neither TBP nor hTAF_II140 was detected in control immunoprecipitations using the anti-Flag antibody (lane 1). Reprobing of the immunoblot shows that previously identified TAF_IIs were present in the immunopurified TFIID as expected (Fig. 3B, lanes 4 to 6). It is possible that TAF_II140 was originally overlooked because it comigrates closely with hTAF_II135 (compare lanes 3 and 4).

The anti-hTAF_II140 antibodies were also used to immunoprecipitate HeLa cell nuclear extracts. TBP, TAF_II30, TAF_II20, TAF_II18, and TAF_II55 as well as other TFIID components (data not shown) were all coprecipitated with TAF_II140 (Fig. 3C, lanes 1 to 4). TAF_II30 is a component of the TFIID and TFTC complexes (5, 53). To determine whether TAF_II140 is also present in TFTC, we asked whether TFTC-specific components could also be immunoprecipitated with the anti-TAF_II140 antibody. Incubation of the immunoblots with antibodies against hGCN5 and SAP130 (Brand et al., submitted) indicated that both were coprecipitated (Fig. 3C). Therefore, the anti-TAF_II140 antibody precipitates both TFIID-specific subunits (TBP, hTAF_II18) and TFTC-specific subunits (SAP130, GCN5), indicating that it is present in both complexes.

BIP2 is a Drosophila TFIID subunit which heterodimerizes specifically with dTAF_II24.

Database searches with mTAF_II140 show that it shares significant sequence similarity with the Drosophila protein BIP2 (Fig. 4A). BIP2 was first identified (J.-C. Pointud and J.-L. Couderc, unpublished data) in a two-hybrid screen using as bait the BTB (POZ) domain of bric-a-brac 1, a regulatory protein required for several morphogenetic processes in Drosophila development (21, 45). BIP2 contains an HFD in the N terminus and a PHD finger at the C terminus (Fig. 4A). The PHD fingers of mouse and human TAF_II140 and BIP2 are more closely related to each other than to PHD fingers from other proteins (Fig. 4B). The significant similarity with hTAF_II140 led us to ask whether BIP2 is a component of dTFIID.

We first determined whether the dBIP2 HFD would heterodimerize with those of dTAF_II16 or dTAF_II24. In yeast two-hybrid assays, the VP16-dBIP2 HFD chimera interacted strongly with the LexA-dTAF_II24 fusion, but only a weak interaction with dTAF_II16 was observed (Fig. 5A, lanes 1 and 2). In contrast, no significant interaction with the HFDs of yTAF_II25 or hTAF_II30 was observed (lanes 3 to 5). This indicates that the dBIP2 HFD can discriminate between the HFDs of dTAF_II16 and dTAF_II24. The selective interaction of the dBIP2 HFD with that of dTAF_II24 was also investigated by coexpression in Escherichia coli cells. GST-fusions of the BIP2 HFD were coexpressed with native dTAF_II16 and dTAF_II24. The GST-BIP2(2-89) fusion was solubilized by coexpression of full-length dTAF_II24, for which a complex of the two was retained on the resin, whereas with dTAF_II16, no solubilization nor heterodimerization was observed (Fig. 5B, lanes 2 to 3). Similarly, GST-BIP2(2-89) was solubilized by coexpression of the dTAF_II24 HFD(57-167), but not by the dTAF_II16 HFD(41-146) (lanes 4 and 5). The GST-BIP2-dTAF_II24 HFD heterodimerization was almost completely abolished by mutation of residues within the L1 loop [lane 8, G23R, Y24K, GST-BIP2(2-89)m1]. These results indicate a direct and selective dBIP2-dTAF_II24 heterodimerization via their HFDs.

FIG. 5.

(A) Graphical representation of quantitative two-hybrid β-galactosidase assays. The VP16-dBIP2 chimera is shown schematically above the graph and interactions with the LexA chimeras shown below each column are indicated. −, negative controls with the LexA domain alone. (B) Direct and selective dBIP2-dTAF_II24 heterodimerization. Bacteria were transformed to express the proteins shown above each lane, and following extract preparation, the soluble protein retained on glutathione-Sepharose beads was analyzed by SDS-PAGE and staining with Coomassie brilliant blue. /, the GST-fusion was expressed alone. For lanes 2 and 3, coexpression was performed with the full-length dTAF_II16 and dTAF_II24 proteins as indicated, while for lanes 4, 5, 7, and 8, coexpression was performed with the HFDs of the indicated proteins. The locations of the GST-BIP2 fusions and the dTAF_II24 proteins are indicated. (C and D) Drosophila embryo nuclear extract was precipitated with antisera against TBP (lane 4) or an irrelevant antiserum (raised against GST) (lane 3) as indicated. Lane 2 shows a control immunoprecipitation with the anti-TBP serum in the absence of added nuclear extract. ∗, locations of the precipitated proteins.

A polyclonal anti-dBIP2 serum was generated (see Materials and Methods). In immunostaining, the anti-dBIP2 serum, but not the preimmune serum, recognized a ubiquitously expressed nuclear protein throughout Drosophila embryogenesis (Fig. 6A and data not shown). This antibody also localized the dBIP2 protein on polytene chromosomes exclusively in decondensed regions, both puffs and interbands (Fig. 6B, arrowheads) which correspond to RNA polymerase II start sites in transcriptionally active chromatin (30, 46). In contrast, it is not found in the transcriptionally inactive heterochromatin around the centromere.

FIG. 6.

(A) Immunodetection of dBIP2 in the nuclei of Drosophila embryos. The relevant stages are indicated. (B) In situ immunodetection of dBIP on salivary gland polytene chromosome spreads. The arrowheads indicate examples of interband regions staining strongly with the anti-dBIP2 serum. Cent., the centromeric region.

To determine whether dBIP2 is indeed a dTFIID subunit as suggested by its ability to interact with dTAF_II24, immunoprecipitation experiments were performed. The anti-dBIP2 serum recognized a protein with a mass of around 155 kDa in Drosophila embryo nuclear extracts (Fig. 5C, lane 1). The embryo extract was precipitated with an antiserum against dTBP, which has previously been shown to precipitate the dTFIID complex (19, 29). BIP2 was precipitated with the anti-dTBP sera (Fig. 5C, lane 4), while no immunoprecipitation was seen using an irrelevant antiserum (lane 3), and the signal did not result from a spurious cross-reaction with the anti-TBP serum (lane 2). As a positive control, dTBP, dTAF_II42, dTAF_II24, and dTAF_II230 were also selectively precipitated by the anti-TBP serum (Fig. 5C and D, lanes 4, and data not shown). These data indicate that dBIP2 (dTAF_II155) can be coimmunoprecipitated with dTBP and other dTAF_IIs and hence is a subunit of dTFIID.

A functional domain of yeast TAF_II65 required for viability in yeast is shared with the Drosophila Prodos protein.

The above results indicate that dBIP2 (dTAF_II155) is a TFIID component which selectively heterodimerizes with dTAF_II24. It has recently been reported that the PDS gene product contains an HFD and interacts specifically with dTAF_II16, leading to the idea that PDS is a TFIID subunit (25). Here, we show that dBIP2 (dTAF_II155) selectively heterodimerizes with dTAF_II24. yTAF_II25 can heterodimerize with either yTAF_II47 or yTAF_II65. Hence, if dTAF_II155 and hTAF_II140 are homologues of yTAF_II47, it is possible that PDS may be a homologue of yTAF_II65. We have previously shown that, in addition to the HFD, a second domain of yTAF_II65 located between amino acids 103 and 510 was essential for its function in yeast (17). As functional domains in TAF_IIs are often evolutionarily conserved, we examined this region for similarities with PDS. Sequence comparisons indicate that yTAF_II65 shares significant sequence similarity with PDS not only in the HFD but also in the region C terminal to the HFD (Fig. 7A). Several highly conserved sequence motifs separated by insertions corresponding to potential loop regions in yTAF_II65 were observed. We therefore wished to determine whether this conserved domain (designated the TAPD) was essential for yTAF_II65 function.

FIG. 7.

(A) Comparison of the amino acid sequences of yTAF_II65, PDS, TBN, and a putative Xenopus TBN as described by Voss et al. (51). Identical amino acids are shown as white letters in a black background, and similar amino acids are boxed in gray. The location of the histone fold is indicated. The endpoints of the deletions in yTAF_II65 are indicated. The highly conserved HTY amino acids mutated in m2 are indicated (∗). (B) The yTAF_II65 deletions are schematized along with their ability to rescue the growth of the yTAF_II65 null strain. The presence of mutations is indicated (∗). The mutated amino acid sequences are shown at the bottom. TS, temperature sensitivity.

Using the plasmid shuffle technique for yeast, we tested the ability of derivatives of yTAF_II65 with a deletion or mutation in the TAPD to complement the growth of a yTAF_II65 null strain. Deletion of the yTAF_II65 HFD does not abolish growth at 30°C but generates a temperature-sensitive phenotype (17), indicating that the region containing the TAPD contributes to yTAF_II65 function. We made further deletions to determine more precisely the location of this functional domain of yTAF_II65.

Deletion of amino acids downstream of 406 did not affect the ability of yTAF_II65 to rescue the growth of the null strain at 30°C, while a C-terminal truncation at amino acid 320 was unable to rescue growth [Fig. 8A, sections 2 to 4, summarized in Fig. 7B, compare yTAF_II65(103-406) and yTAF_II65(103-320)]. At the N terminus, a derivative beginning at amino acid 161 was able to complement growth, but a further deletion up to position 192, which deletes a highly conserved region of the TAPD, abolished function [Fig. 8A, sections 5 and 6, summarized in Fig. 7B, compare yTAF_II65(161-406) and yTAF_II65(192-510)]. As all of the viable constructs lacked the HFD, they showed a temperature-sensitive phenotype, being unable to complement at 37°C (data not shown). Therefore, a minimal TAPD between amino acids 161 and 406 is required for yTAF_II65 function.

FIG. 8.

The growth of yeast plated at the indicated temperatures is shown. The complementing plasmids used are next to the plates. WT, wild type.

The minimal domain described above comprises the most conserved regions of the TAPD. To further test the significance of the sequence similarity and to evaluate the relative contributions of the HFD and TAPD to yTAF_II65 function, we deleted or mutated one of the most conserved blocks between amino acids 170 and 200 in the context of wild-type yTAF_II65 and in the context of yTAF_II65 m1, where the HFD is mutated. When this region was deleted in the context of a wild-type HFD, complementation was observed [Fig. 8B, section 2, summarized in Fig. 7B, yTAF_II65(Δ170-200)]. In contrast, when the deletion was made in the presence of the mutated HFD, which by itself is viable at 30°C (Fig. 8B, section 1), no growth was observed [Fig. 8B, section 3, compare yTAF_II65(1-510)m1 with yTAF_II65(1-510)m1Δ]. Similarly, mutation of the highly conserved HTY residues within the region had no effect in the presence of the wild-type HFD but led to a loss of function in the context of a mutated HFD [Fig. 8B, sections 4 and 5, compare yTAF_II65(1-510)m2 and yTAF_II65(1-510)m3, summarized in Fig. 7B]. Hence, when the HFD is functional, the TAPD can be mutated, and likewise, when the TAPD is intact, the HFD is not absolutely required. These results indicate that yTAF_II65 contains two partially redundant domains. Nevertheless, the HFD seems to play a more important role since its mutation leads to a temperature-sensitive phenotype while mutation or deletion of the TAPD alone does not [compare yTAF_II65m1, yTAF_II65m2, and yTAF_II65(Δ170-200) in Fig. 8C and D].

DISCUSSION

TAF_II140 and BIP2 are novel metazoan TFIID subunits.

We describe here mouse and human TAF_II140 and dBIP2 (dTAF_II155), novel components of the mammalian and Drosophila TFIID complexes. Both of these proteins contain an N-terminal HFD which selectively heterodimerizes with that of members of the yTAF_II25 family. The mouse and human TAF_II140 HFD heterodimerizes with those of both yTAF_II25 and hTAF_II30, whereas the dTAF_II155 HFD is much more selective, interacting only with that of dTAF_II24 and not with that of dTAF_II16. Most other organisms seem to have only one homologue of yTAF_II25. The existence of two closely related genes in Drosophila has allowed each one to evolve a more specialized role. For example, as previously reported (19), dTAF_II24 is associated with both the TFIID and the GCN5-containing complexes, while dTAF_II16 is present only in the TFIID complex. The differential presence of these proteins in the TFIID and GCN5-containing complexes imposes a specialization in partner choice. While yTAF_II47 and mouse and human TAF_II140 can heterodimerize with yTAF_II25 or hTAF_II30, the dTAF_II155 and PDS HFDs are much less promiscuous in their partners since there is a selective interaction of PDS with dTAF_II16 (25) and dTAF_II155 with dTAF_II24. Consequently, the specialization of the dTAF_II16 and dTAF_II24 HFDs has forced the dTAF_II155 and PDS HFDs to also evolve towards a restricted partner choice. This restriction contributes to the relative subunit compositions of the TFIID and GCN5-containing complexes.

TAF_II140 and dTAF_II155 also contain a PHD finger at the C terminus of the protein. The PHD finger is found in almost 300 proteins, most of which are nuclear (39). The majority of these proteins are involved in transcription and interactions with chromatin, such as the Polycomb-like family, and more recently the ATP-dependent chromatin remodeling factor ACF (1, 41). Mutations in the PHD fingers of several proteins have been associated with disease (for examples, see references 4 and 20). The structures of the PHD finger of the Williams-Beuren syndrome transcription factor (WSTF) and the KAP-1 corepressor have been recently solved by nuclear magnetic resonance and shown to comprise an interleaved zinc finger which binds two zinc ions (9, 39).

While the function of the PHD finger is unknown, it has been suggested that it is involved in protein-protein interactions. This is the first time a PHD finger has been found in a TFIID subunit, and it may mediate TFIID interactions with chromatin or with chromatin-associated proteins. Its presence in TFIID is particularly intriguing since many of the factors in which it is found are associated with repressor activity, histone deacetylases, and interactions with heterochromatin (for an example, see references 15, 31, and 55). BIP2, however, is not localized on heterochromatin but is associated with transcriptionally active regions of polytene chromosomes. The PHD finger is not the only motif characteristic of chromatin-interacting proteins to be found in TFIID. TAF_II250 contains bromodomains, also found in many chromatin-interacting factors, which have recently been shown to mediate interaction with acetyl-lysine groups in the histone tails (12, 26). Hence, both the bromodomains and the PHD finger may be interfaces for directing TFIID to promoters within chromatin.

Immunoprecipitations demonstrate that both hTAF_II140 and dTAF_II155 are TFIID subunits, since they are coprecipitated under stringent conditions along with TBP and TAF_IIs. It has been previously shown that dTAF_II24 is coprecipitated by the dTAF_II16 antisera and vice versa (19). The simplest interpretation is that both dTAF_II24 and dTAF_II16 are present in the same dTFIID complex containing one (or more) molecules of each. This TFIID would therefore contain both PDS and dTAF_II155. An analogous result was obtained with yeast, where yTAF_II47 can be precipitated from strains expressing a tagged version of yTAF_II65 (45). These data were supported by immunoelectron microscopy which showed the presence of multiple molecules of yTAF_II65, yTAF_II47, and yTAF_II25 in the same TFIID particle (C. Leurent, S. Sanders, V. Mallouh, P. A. Weil, D. B. Kirschner, L. Tora, and P. Schultz, submitted for publication).

In HeLa cells, hTAF_II30 is a component of both TFIID and TFTC. When HeLa cell nuclear extracts were immunoprecipitated with the anti-TAF_II140 monoclonal antibody, both TFIID-specific components, such as TAF_II18 and TBP, and TFTC-specific components, such as GCN5 and SAP130 (also ADA3 and SPT3; our unpublished data), were detected. The TAF_II30-TAF_II140 heterodimer therefore appears to be a component of both TFIID and TFTC. At first sight this may seem surprising since, in yeast, yTAF_II47 is TFIID specific and is not present in SAGA where the SAGA-specific ySPT7 heterodimerizes with yTAF_II25 (17). However, hTAF_II135 and hTAF_II150 are TFTC components, yet their yeast homologues, yTAF_II48 and yTAF_II150, are not present in SAGA. In this respect, the findings for yeast and mammalian cells are somewhat different and reflect the fact that TFTC and SAGA have similar, but not identical, compositions. The observation that hTAF_II140 is a TFTC component does not, however, rule out the existence of a hSPT7 or dSPT7 homologue, as we previously suggested (17).

The evolutionarily conserved TAPD is required for yTAF_II65 function in vivo.

Here, we describe the novel TAPD present in yTAF_II65 and PDS. The presence of an HFD and the TAPD in PDS which is known to interact physically and functionally with dTFIID components (25) indicates that PDS is a yTAF_II65 homologue. These domains are also present in the mammalian protein TBN, further suggesting that it may also be a TAF_II (25, 51).

The TAP domain is located C terminal of the HFD, and in yTAF_II65 it is partially redundant with the HFD. In complementation experiments, deletion or mutation of the HFD does not abolish yTAF_II65 function when the TAPD is wild type. Similarly, the TAPD can be mutated or deleted without loss of function when the HFD is present. In contrast, simultaneous mutation of both leads to a loss of function. Hence, the HFD and the TAPD play redundant roles in yTAF_II65 function.

As the function of the yTAF_II65 HFD is to mediate heterodimerization with yTAF_II25 and hence the integration of yTAF_II65 into TFIID, it is likely that the TAPD is also an interface for interactions between yTAF_II65 and another TAF_II(s). In the absence of the HFD, the TAPD would allow yTAF_II65 to associate with TFIID and support growth and vice versa. Obviously, when both are mutated, it is no longer possible for yTAF_II65 to interact with the TFIID, and this is lethal for yeast. Note, however, that the HFD seems to be the more critical for yTAF_II65 function, since all the viable strains in which it was deleted had a temperature-sensitive phenotype, whereas the strains in which only the TAPD was mutated grow at 37°C. This is consistent with the fact that histone-like heterodimers are extremely stable and suggests that the interactions mediated by the TAPD are less so. Nonetheless, it is worth noting that the TAPD comprises several conserved sequence blocks, suggesting that it may be composite in nature. Thus, even when amino acids 170 to 200 are deleted or mutated, the remainder of the domain may act cooperatively with the HFD to allow growth at 37°C. We have previously reported that the temperature sensitivity resulting from mutation or deletion of the yTAF_II65 HFD could be selectively suppressed by overexpression of yTAF_II68 (17). This suggests that the TAPD may interact with yTAF_II68. Biochemical and structural analyses will be required to determine how the TAPD interacts with yTAF_II68 and/or other TFIID components.

In this study, we have proposed that BIP2 (dTAF_II155) and hTAF_II140 are homologues of yTAF_II47 due to the similarity in sequence of the HFDs and their ability to heterodimerize with the yTAF_II25 homologues in dTFIID and hTFIID and that PDS is a homologue of yTAF_II65 based on the shared HFD and TAPD. These results indicate a remarkable conservation of TFIID subunit composition. One notable exception to this is the presence of a PHD finger in hTAF_II140 and its absence in yTAF_II47. Nevertheless, in yeast, the bromodomains and kinase activity in hTAF_II250 are not present in yTAF_II130 (yTAF_II145) but are provided by the association of the Bdf1 and Bdf2 proteins with TFIID (34). Database searches with the mTAF_II140 and hTAF_II140 PHD finger revealed the presence of several yeast proteins with related PHD fingers. It is possible that, by analogy with Bdf1 and Bdf2, one of these PHD finger-containing proteins associates with yTFIID to provide the missing domain.