Identification of a transactivation function in the progesterone receptor that interacts with the TAFII110 subunit of the TFIID complex.

Transcriptional activation of target genes by the human progesterone receptor is thought to involve direct or indirect protein-protein interactions between the progesterone receptor and general transcription factors. A key role in transcription plays the general transcription factor TFIID, a multiprotein complex consisting of the TATA-binding protein and several tightly associated factors (TAFs). TAFs have been shown to be required for activated transcription and are, thus, potential targets of activator proteins. Using in vitro interaction assays, we could identify specific interactions between the progesterone receptor and the TATA-binding protein-associated factor dTAFII110. The dTAFII110 domain responsible for the interaction is distinct from that reported to suffice for binding to Sp1. Somewhat surprisingly, deletion analysis indicated that the previously identified activation functions 1 and 2 of the progesterone receptor are not required for this interaction but pointed to an important role of the DNA binding domain. In cotransfection experiments and an in vitro transcription assay, the DNA binding domain of the progesterone receptor displayed significant activation potential. These findings, taken together, suggest that an interaction between the progesterone receptor and TAFII110 may represent an important step in the mechanism of activation.

In eukaryotes, at least seven basal transcription factors (TFIIA, TFIIB, TFIID, TFIIE, TFIIF, TFIIH, and TFIIJ) are required for basal levels of transcription at promoters by RNA polymerase II in vitro (reviewed in Ref. 1). In mammalian systems, these factors assemble into functional initiation complexes in a highly ordered, stepwise fashion beginning with the binding of TFIID to the TATA box (1)(2)(3), while in yeast a number of factors seem to be brought in as a preassembled RNA polymerase II holoenzyme complex (4). Sequence-specific activators that bind to specific regulatory elements within distal promoter regions or enhancers are capable of stimulating the rate of transcription initiation. Although the precise mechanism is presently unknown, there is increasing support for models proposing that protein-protein interactions between activators and basal transcription factors play a decisive role in this process (5). Such interactions might help to recruit these basal transcription factors or the preassembled RNA polymerase II holoenzyme complex to the TATA box, stabilize intermediates in the assembly of the initiation complex, or activate certain components by inducing conformational changes.
Biochemical and molecular characterization of TFIID has shown that it is a multi-subunit complex consisting of the TATA-binding protein (TBP) 1 and a number of TBP-associated factors (TAFs) (6 -8), ranging from 20 to 250 kDa in size. In cell-free transcription systems reconstituted from partially purified factors, recombinant TBP is able to support basal transcription of TATA-only promoters (9,10). However, unlike TFIID, TBP fails to mediate transcriptional stimulation by various activators, indicating that TAFs are required for this process (8,10). Consequently, a model has been proposed, suggesting that different activators might interact with different TAFs in the TFIID complex (3,7). Indeed, interactions with subunits of Drosophila (11)(12)(13)(14)(15)(16) and human (15,17,18) TFIID complexes have been demonstrated for a growing number of transcription activators.
The progesterone receptor (PR) is a ligand-inducible transactivator that belongs to the superfamily of nuclear receptors (19 -21). Activated PR facilitates the formation of initiation complexes through binding to progesterone response elements (PREs) located in the promoter region of target genes (22). Molecular analysis has shown that PR, like other transcription activators, consists of separable domains responsible for DNA binding (DBD) and transcriptional activation (23). The aminoterminal activation function (AF-1) is constitutively active, whereas the activation function located within the carboxylterminal part (AF-2) requires hormone for its activity (24,25).
In the present study, possible interactions of hPR with a subunit of the TFIID complex were examined by protein-protein interaction assays. Our results show that the 110-kDa subunit of Drosophila TFIID (dTAF II 110) is specifically bound by hPR in vitro and that this interaction is mediated by specific domains of both proteins. Transfection studies and in vitro transcription experiments revealed that the part of hPR which mediates the interaction contains a previously unidentified activation function.

MATERIALS AND METHODS
Baculovirus Expression-cDNAs encoding full-length hPR (form B) and deletions were derived from expression vector hPR0 (26). By insertion into a Bluescript SK ϩ derivative encoding the peptide MSHHHH-HHTSETY, 2 the hPR cDNAs were fused in frame to a His-tag. Two NheI sites flanking the cDNAs were used to release the tagged fragments by NheI digest. To generate baculovirus transfer vectors, these NheI fragments were ligated into the NheI site of pBlueBac (Invitrogen). In the transfer vector encoding MH 6 -hPR0, the His-tag is followed by aa 1-933 of the hPR0 open reading frame (26), in MH 6 -hPR0⌬core by aa 1-456/538 -933, in MH 6 -ABC by aa 1-639 followed by Arg, in MH 6 -ABC⌬core by aa 1-456/538 -639 followed by Arg, in MH 6 -BC by aa 165-639 followed by Arg, in MH 6 -BC⌬core by aa 165-456/538 -639 followed by arg, in MH 6 -C by aa 538 -639 followed by Arg, in MH 6 -AB⌬core by aa 1-456/538 -555 followed by Gln-Asn-Ser, and in MH 6 -B⌬core by aa 165-456/538 -555 followed by Gln-Asn-Ser.
Similarly, a cDNA encoding a His-tagged hTBP was constructed, starting from the expression vector pTM-hTFIID (9). In the final pBlue-Bac derivative, the first ATG codon of the hTBP cDNA was preceded by a DNA sequence encoding MSHHHHHHTSGSPSLI.
Transfer vectors were cotransfected with linearized baculovirus DNA (Pharmingen) into Sf21 cells. Baculoviruses expressing the desired proteins were identified by SDS-PAGE, Western blotting with appropriate antibodies, or electrophoretic mobility shift assay with extracts derived from cells infected with plaque-purified recombinant viruses.
Expression in Escherichia coli and Protein Purification-Construction of an expression vector encoding aa 536 -664 of hPR as His-tagged fusion protein was performed by a PCR-based approach using MH 6 -hPR0 (see below) as template and appropriate primers containing 5Јand 3Ј-BamHI restriction sites. The PCR fragment was inserted into BamHI-cut pQE-30 (Qiagen) creating the vector pQE-hPR(DBD), which encodes the pQE His-tag followed by Gly-Ser-Thr-Ser and aa 536 -664 of hPR.
For protein expression, a 250-ml culture of E. coli TG1 containing pQE-hPR(DBD) was grown at 37°C in TB medium containing 100 g/ml ampicillin to an A 595 of 0.6. Expression was then induced by addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM. Cells were harvested 4 h later. Bacterial pellets were resuspended in 10 ml of NaP buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 2 mM ␤-mercaptoethanol). Cells were lysed by sonication, and insoluble material was removed by centrifugation for 30 min and 13.000 rpm in a Beckmann JA20 rotor. To purify the MH 6 -hPR(DBD) protein, 100 l of a 50% slurry of Ni-NTA resin (Qiagen) equilibrated in NaP buffer was added following incubation at 4°C for 1 h. Subsequent washing steps included washing with NaP buffer, buffer D100 (20% glycerol, 100 mM KCl, 25 mM HEPES, pH 7.9, 10 mM ␤-mercaptoethanol), buffer D100 plus 5 mM imidazole, and buffer D100 plus 20 mM imidazole. MH 6 -hPR(DBD) was eluted with buffer D100 plus 100 mM imidazole and 40 nM EGTA, dialyzed against 20 mM HEPES, pH 7.6, 50 mM KCl, 10% glycerol, 0.1 mM EGTA, 1.5 mM MgCl 2 , 1 M ZnSO 4 , 2 mM dithiothreitol and stored at Ϫ80°C. All buffers contained protease inhibitors. Preparation of an E. coli mock extract was performed following the same protocol, except that untransformed E. coli TG1 cells were grown in media lacking ampicillin.
Expression Vectors for in Vitro Translation-Expression vectors for in vitro transcription/translation of dTAF II 110 and mutants were derived from pT␤110 (kindly provided by T. Hoey and R. Tjian). To construct carboxyl-terminal deletions of dTAF II 110, different parts of the cDNA were removed from pT␤110 (12) by double digestion with BssHII, which cuts within the 3Ј-untranslated region, and appropriate restriction enzymes within the coding region of dTAF II 110. After repairing the ends with Klenow enzyme, the fragments containing the vector and amino-terminal sequences of dTAF II 110 were isolated and religated. Stop codons in all three reading frames were provided by a sequence located immediately downstream of the original BssHII site. The following constructs correspond to the following restriction sites within the coding region: TAF1-787, XhoI; TAF1-684, HpaI; TAF1-310, SalI; TAF1-137, ClaI.
Construction of expression vectors encoding amino-terminally de- In Vitro Protein-Protein Interaction Assays-For in vitro proteinprotein interaction assays, [ 35 S]methionine-labeled proteins were synthesized in rabbit reticulocyte lysate using a coupled in vitro transcription/translation system (TNT, Promega) as described by the manufacturer.
Interaction assays using oligonucleotide-bound hPR were performed basically as described by Hoey et al. (12). 5Ј-Biotinylated DNA fragments containing two PREs and a TATA box were synthesized by PCR using the vector PRE 2 TATA-CAT (see below) as template and the oligonucleotides Biotin-TGTAAAACGACGGCCAG and CGAATTCCT-GCAGCCC as primers. Binding of the 171-base pair fragments to streptavidin-agarose beads (Pierce) was performed by overnight incubation at 4°C in buffer T (25 mM HEPES, pH 7.6, 1 mM EDTA, 0.01 mM ZnSO 4 , 2 mM dithiothreitol, 10% glycerol, 0.01% Nonidet P-40) containing 0.1 M NaCl (buffer T 0.1 M) with subsequent removal of unbound DNA by three washing steps. 15 l of beads (50% suspension) lacking or containing 1 g of DNA were incubated in the absence or presence of 5 g of partially purified MH 6  For interaction assays using proteins directly immobilized to a matrix, the His-tagged proteins were bound to Ni-NTA-agarose beads (Qiagen, 10 l of a 50% suspension) by incubation in buffer T 0.1 M for 1 h at 4°C. Washing of the beads, incubation with [ 35 S]methioninelabeled proteins, and analysis were done as described above.
SDS-PAGE and Silver Staining-SDS-PAGE and silver staining were performed as described by Sambrook et al. (28).
Transient Transfection and CAT Assay-The reporter plasmids PRE 2 TATA-CAT and TATA-CAT were constructed by replacing the SacI-SalI fragment containing the G-free cassettes of PRE 2 TATA-G300 and TATA-G400 (29), respectively, with a SacI-SalI fragment from BS-CAT 2 containing a CAT cassette. BS-CAT had been constructed by inserting the HindIII fragment of pCAT promoter (Promega) into the HindIII site of Bluescript SK ϩ so that the 5Ј-end of the CAT gene was flanked by the SacI site of the multiple cloning site.
To construct the eukaryotic expression vector pSGN-MH 6 -hPR0, the His-tagged cDNA was excised by NheI restriction from the corresponding pBlueBac transfer vector (see above) and inserted into the NheI site of pSGN. pSGN had been derived from pSG5 (30) by inserting an NheI 8-mer linker into the refilled BglII site. pSGN-hPR(C) encoded a methionine residue followed by aa 556 -639 of hPR0, and ARG and was made by a PCR-based approach. Correct reading frames and the absence of mutations in PCR-amplified fragments were verified by automated sequencing (ALF DNA-Sequencer, Pharmacia Biotech Inc.) with appropriate primers.
Cos7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics. One day prior to transfection, 10 5 cells were seeded onto 6-cm dishes. Transfections were performed using the Lipofectin method according to the manufacturer's (Life Technologies, Inc.) recommendation. Briefly, reporter DNA (2 g) 2 L. Klein-Hitpass, unpublished data. and expression vector (100 ng) were diluted into 100 l of OPTIMEM (Life Technologies, Inc.) and made up to a total of 2.5 g of DNA by the addition of empty expression vector. This solution was mixed with an equal volume of OPTIMEM containing 6 g of Lipofectin. 15 min after mixing, 1.8 ml of OPTIMEM was added, and the liposome/DNA mixture was poured onto the cells, which had been washed with phosphatebuffered saline and serum-free OPTIMEM. After 4 h, the solution was removed, and 3 ml of phenol red-free Dulbecco's modified Eagle's medium containing 10% fetal calf serum was added. Progesterone treatment was started by adding 3 l of a 0.1 mM stock in ethanol; controls received 3 l of ethanol. Cells were harvested 40 h after transfection. Preparation of cell extracts and CAT assays were done as described (31).
In Vitro Transcription-In vitro transcription reactions contained 6 l of rat liver nuclear extract (6.3 g/l) prepared as described by Döbbeling et al. (32). E. coli mock extract, recombinant MH 6 -hPR(DBD), and recombinant MH 6 -hPR0 added to some of the reactions were expressed in E. coli or Sf21 cells infected with recombinant baculovirus, respectively, and purified as described above. The supercoiled transcription templates PRE 2 TATA-G300 and TATA-G400 containing G-free cassettes (33), reaction conditions, and analysis of transcripts were as previously described (29,34,35). Quantitation of transcripts was performed by autoradiography and subsequent densitometry.

RESULTS
hPR Interacts Specifically with dTAF II 110 in Vitro-It has been shown that various activators can interact through their activation domains with TAFs (11)(12)(13)(14)(15)(16)(17)(18). In vitro transcription interference experiments had indicated that steroid receptors and Sp1 might have a common target(s) in the transcription machinery (29). Combined with the report of Hoey et al. (12), which identified dTAF II 110 as a target of Sp1, these findings thus raised the possibility that hPR and dTAF II 110 might interact with each other. To address this question, we performed protein-protein interaction assays with [ 35 S]methionine-labeled dTAF II 110 and the baculovirus expressed Histagged hPR (MH 6 -hPR0). In initial experiments, partially purified MH 6 -hPR0 was bound to PRE-containing DNA fragments, which were immobilized to streptavidin-agarose beads via a biotin moiety. After incubation with labeled dTAF II 110, unbound protein was removed by washing with binding buffer, and bound proteins were eluted with buffer containing 1 M NaCl, separated by SDS-PAGE, and visualized by autoradiography. As shown in Fig. 1A, 35 S-labeled dTAF II 110 was retained on beads containing DNA-bound MH 6 -hPR0, whereas no binding was observed on DNA beads lacking MH 6 -hPR0 (compare lanes 6 and 5), demonstrating that dTAF II 110 does not stably bind to DNA on its own. In contrast, control beads lacking DNA fragments, which did not contain bound PR after washing (data not shown), did not bind dTAF II 110 (lanes 4 and 3). To explore the specificity of the interaction, we tested whether DNA-bound MH 6 -hPR0 would interact under the same conditions with 35 S-labeled luciferase (Fig. 1B) or human TFIIB (Fig. 1C) and found that both proteins were not retained by the receptor. These results demonstrate that the interaction of DNA-bound hPR with dTAF II 110 is specific.
The sensitivity of the assay involving DNA-bound hPR was relatively low, probably due to gradual dissociation of hPR-dTAF II 110 complexes from the DNA during the washing steps. We thus analyzed whether the interaction would also occur with MH 6 -hPR0 immobilized directly to Ni-NTA-agarose beads through the amino-terminal His-tag. As shown in Fig. 2A, dTAF II 110 was indeed bound on beads loaded with MH 6 -hPR0 (lane 4) but not on control beads without MH 6 -hPR0 (lane 3). Furthermore, a His-tagged fusion protein representing XDCoH (kindly provided by E. Pogge v. Strandmann and G.U. Ryffel), the Xenopus homologue of DCoH (36), did not bind dTAF II 110 (Fig. 2B). These results confirm that specific interactions between hPR and dTAF II 110 can also occur with hPR immobilized to Ni-NTA-agarose beads. Even though the interaction could be observed with hPR immobilized to Ni-NTA-agarose, an involvement of DNA could not be excluded as in vitro transcription/translation lysate programmed with a significant amount of dTAF II 110 expression plasmid was used in the experiment. Therefore, the interaction assay was performed in the presence of increasing amounts of the DNA intercalator ethidium bromide (EtBr), which inhibits DNA binding and thus allows to discriminate between DNAdependent and DNA-independent protein interactions (37). As shown in Fig. 2A, EtBr up to 100 g/ml (lanes 5 and 6), a concentration shown to inhibit DNA-dependent protein associations (37), had little effect on the interaction, while at relatively high concentrations of 200 and 400 g/ml (lanes 7 and 8) significant inhibition was obtained. However, even at the highest EtBr concentration used, the association of hPR and dTAF II 110 could not be completely inhibited. The partial sensitivity to EtBr suggests that there is a DNA-independent component reflecting bona fide protein-protein interactions, which are possibly further stabilized through DNA.
Amino acids 538 -639 of hPR Are Sufficient for Interaction with dTAF II 110-As has been demonstrated by the groups of Gronemeyer and Chambon (24,25), the hPR contains two activation functions (AF-1 and AF-2) located amino-and carboxyl-terminally of homology region C, which comprises the DBD (see Fig. 3A). AF-1 has been further dissected into a core region, which mediates transactivation when linked to a GAL4-DBD and a modulatory domain, which exhibits activation potential only when fused to the homologous DBD (38,39). To determine which domain of the receptor mediates the binding to dTAF II 110, we expressed a number of hPR mutants lacking either single or combinations of domains as His-tagged fusion proteins in insect cells (Fig. 3A). The purity of the various receptor proteins after affinity purification on Ni-NTA-agarose was examined by SDS-PAGE and silver staining (Fig. 3B). In addition to the full-size products, MH 6 -hPR0 and MH 6 -hPR0⌬core showed a number of smaller polypeptides (lanes 2 and 3), most of which could be identified as carboxyl-terminally truncated hPR molecules by Western blotting with appropriate antibodies (data not shown). In contrast, all proteins lacking the carboxyl-terminal AF-2 (MH 6 -ABC, MH 6 -ABC⌬core, MH 6 -BC, MH 6 -BC⌬core, MH 6 -C, MH 6 -AB⌬core, MH 6 -B⌬core) were isolated as relatively pure full-size products (lanes 4 -10).
To analyze the interaction potential of the various hPR mutants, we employed the assay in which the receptor proteins were immobilized to Ni-NTA-agarose because of the advantage that one can also examine mutants, which do not contain a DBD (MH 6 -AB⌬core and MH 6 -B⌬core). Using this set of mutants, we could demonstrate that progressive deletion of AF-1 and AF-2 sequences up to a construct containing only amino acids 538 -555 of AF-1 and amino acids 556 -639 of the DBD (MH 6 -C) did not eliminate binding of dTAF II 110 (Fig. 4A). In contrast, two different deletion mutants lacking domain C but containing amino acids 538 -555 of AF-1 (MH 6 -AB⌬core and MH 6 -B⌬core) showed negligible binding of dTAF II 110, even when 6-fold higher amounts of the receptor proteins were tested (Fig. 4B). Thus, our experiments show that homology region C containing the DNA binding domain plays an impor-  5-8), EtBr was included in the binding reaction. The position of bound dTAF II 110 is indicated on the right side of the autoradiograph. B, same as A, except that 5 g of His-tagged XDCoH were immobilized on the beads. Lane 1 contains 25% of the dTAF II 110 input.

FIG. 3. Expression and purification of hPR and deletion mutants as His-tagged fusion proteins in insect cells.
A, at the top a schematic representation of the domain structure of the hPR (form B) is given. Homology regions A/B to E are indicated according to Meyer et al. (38). Below, the structures of the His-tagged fusion proteins are shown. The hatched box represents the core activating region of AF-1 (38). The His-tag at the amino-terminal end of the proteins is depicted as a filled box. At the right side of the panel, the dTAF II 110-binding potential of the different constructs is shown. B, SDS-PAGE analysis of the recombinant PR proteins. Proteins were prepared from baculovirus-infected cells as described under "Materials and Methods." 5 g of each protein preparation as determined by the Bradford assay were separated by SDS-PAGE and visualized by silver staining. Molecular weights of marker proteins (lane 1) are given. tant role in the binding of dTAF II 110, while AF-2 and the core and modulatory regions of AF-1 are clearly not required. Minor variations in the level of dTAF II 110 binding observed with the different receptor proteins are likely to be due to different degrees of purity or, in the case of MH 6 -C, due to an overestimation of the protein content by the protein determination method (see Fig. 3B).
Amino Acids 373-787 of dTAF II 110 Are Sufficient for Interaction with hPR-To identify dTAF II 110 sequences important for interaction, we analyzed the association of various deletion mutants of dTAF II 110 with hPR immobilized to Ni-NTA-agarose beads. The structure of the constructs is depicted in Fig. 5A. SDS-PAGE and autoradiography of the various radioactively labeled dTAF II 110 deletions confirmed that all truncated proteins were synthesized with similar efficiency and as apparent full-length proteins (data not shown). Deletion mutants lacking carboxyl-terminal sequences including amino acid 788 (TAF1-787) or 685 (TAF1-684) bound MH 6 -hPR0 almost as efficiently as wild type TAF1-921 (Fig. 5B, compare lanes 4 and 6 with  lane 2). Interestingly, a deletion mutant consisting of amino acids 1-310 (TAF1-310) showed increased nonspecific binding but no preferential interaction with MH 6 -hPR0-loaded beads (lanes 7 and 8). Similarly, the amino-terminal 137 amino acids of dTAF II 110 (TAF1-137) showed no binding to MH 6 -hPR0 (lane 10). Together, these results show that the carboxyl-terminal border of the hPR interaction domain maps between amino acids 684 and 310.
In addition, a deletion construct lacking both amino-and carboxyl-terminal sequences (TAF373-787) was tested for its interaction with MH 6 -hPR0 and various hPR deletion constructs. As shown in Fig. 6, TAF373-787 bound efficiently to MH 6 -hPR0 (lane 4) and all of the hPR deletion constructs shown to be capable of binding TAF1-921, including MH 6 -C (lane 10). Therefore, carboxyl-terminal sequences spanning amino acids 373-787 of dTAF II 110 are sufficient for specific interaction with amino acids 538 -639 of hPR.
Sequences Comprising the dTAF II 110 Interaction Domain of hPR Mediate Transactivation in Vivo and in Vitro-Our conclusion that the DNA binding domain of hPR interacts with a TAF prompted us to investigate whether this part of hPR might be sufficient for transcriptional stimulation. To analyze transcriptional stimulation in vivo, an SV40 early promoter-driven expression vector encoding amino acids 556 -639 of hPR (hPR(C)) was cotransfected with a CAT reporter construct containing two PREs in front of a TATA box (PRE 2 TATA-CAT) into Cos7 cells. Interestingly, expression of hPR(C) stimulated CAT activity to about 10% of the activity observed with MH 6 -hPR0 in the presence of progesterone (Fig. 7A). Transactivation by hPR(C) did not require progesterone treatment (Fig. 7A), as expected, but proved to depend on the presence of PREs in the reporter, as no stimulation was observed with a reporter lacking PREs (TATA-CAT, data not shown).
For analysis of the transcriptional activation potential of the DBD of the hPR in vitro, we used a cell-free transcription system, which uses rat liver nuclear extract as a source for general transcription factors (29,34,35). Two different test genes containing a TATA box or two PREs and a TATA box in front of a G-free cassette (33) were transcribed simultaneously in all reactions. To determine fold activations, correctly initiated transcripts from PRE 2 TATA were normalized to the amount of transcripts obtained from the template containing the TATA box only. As the DNA binding activity in vitro of the E. coli-expressed His-tagged amino acids 556 -664 of hPR (MH 6 -hPR(DBD)) purified by Ni-NTA chromatography was higher than of the Sf21-expressed amino acids 538 -639 (MH 6 -C), we decided to use MH 6 -hPR(DBD) in our in vitro transcription experiments. Amino acids 556 -664 of hPR are homologous to the minimal deletion mutant of the human estrogen receptor, which is sufficient for stable binding to DNA in vitro (40). As shown in Fig. 7B, addition of recombinant MH 6 -hPR0 activated transcription approximately 15-fold, whereas addition of an E. coli mock extract had no effect. Saturating amounts of MH 6 -hPR(DBD) activated transcription to about 13% of the activity observed with MH 6 -hPR0, which agrees well with the value obtained in the transfection experiments. Together, these  6 and 7), or MH 6 -B⌬core (lanes 8 and 9) were used. data prove that the DBD of hPR contains significant transactivation potential. DISCUSSION To detect the interactions between hPR and dTAF II 110 in vitro and to define the domains involved, we used partially purified hPR proteins bound to a Ni-NTA matrix and in vitro synthesized dTAF II 110. In this assay, EtBr caused a clear inhibition (Fig. 2), indicating that DNA is somehow involved. However, since dTAF II 110 alone does not stably bind to DNA (Fig. 1A, lane 5), we exclude the possibility that the association of dTAF II 110 with hPR loaded beads is simply due to capturing of DNA molecules containing bound dTAF II 110 through nonspecific interactions with the DNA binding domain of hPR. Furthermore, we are convinced that the binding of dTAF II 110 to the Ni-NTA matrix is due to an interaction with the hPR and not mediated by a natural poly-histidine containing protein, for example TFIIA (41,42), either present in the reticulocyte lysate or copurified with the PR proteins from infected Sf21 cells for the following reasons. First, we were able to detect hPR-dependent dTAF II 110 binding also in a different assay, in which the hPR was bound to biotinylated DNA fragments immobilized to streptavidin beads (Fig. 1A). Second, the PR dependence of dTAF II 110 binding to Ni-NTA-agarose beads (Figs. 2A, 4, 5, and 6) rules out the possibility that dTAF II 110 is directly bound by a natural poly-histidine containing protein present in the reticulocyte lysate used to synthesize the labeled dTAF II 110 proteins. Third, the negative results obtained with two different hPR mutants (MH 6 -AB⌬core, MH 6 -B⌬core, Fig.  4B), purified from infected Sf21 cells by our standard procedure, strongly argue against copurified poly-histidine containing Spodoptora proteins as being responsible for dTAF II 110 binding. Furthermore, in a chromatographic analysis of a MH 6 -C preparation on a Superose 12 gel filtration column, only one peak of binding activity was observed, which perfectly coincided with the single MH 6 -C polypeptide peak at 13 kDa (data not shown). However, since the dTAF II 110 proteins used in this study were assayed without further purification, we cannot exclude a model in which the hPR-dTAF II 110 interaction is mediated or stabilized by an additional protein(s) present in the translation lysate. Highly purified recombinant dTAF II 110 will be required to investigate this issue.
Prior to this study, two activators have been shown to interact with dTAF II 110, namely Sp1 and CREB (11,12,43). Although the interaction between CREB and dTAF II 110 has not been mapped to a particular region of dTAF II 110 (11), sequence similarities of CREB and Sp1 activation domains suggest that these activators may target the same domain of dTAF II 110 (43). As indicated in Fig. 5A, work of Tjian's group (12) has shown that the amino-terminal 308 amino acids of dTAF II 110 suffice for interaction with Sp1. In contrast, our experiments establish that the interaction with hPR does not involve the amino-terminal region but depends on a more carboxyl-terminal nonoverlapping domain of dTAF II 110. Together with the results of the Sp1-dTAF II 110 interaction, this study identifies dTAF II 110 as the first TAF containing at least two distinct FIG. 5. Mapping of the dTAF II 110 domain required for interaction with MH 6 -hPR0 and MH 6 -C. A, domain structure of wild type TAF II 110 (TAF1-921(wt)) and various deletion constructs. The struc-tural motifs of dTAF II 110 (hatched, S/T-rich; filled, Q-rich; stippled, charged) are shown as described by Hoey et al. (12). Restriction sites used for creating carboxyl-terminal deletion constructs are indicated at the top. At the right side of the panel, the potential of the different constructs for interaction with MH 6 -hPR0 and MH 6 -C is summarized. The regions of dTAF II 110 shown to be sufficient for interaction with Sp1A and Sp1B (12) and hPR and dTAF II 30␣ (48)  interaction surfaces mediating contacts with activators. Due to the multiplicity of activators and the limited number of TAFs, we expect that multiple interaction domains within TAF polypeptides will turn out as a more general feature of these important molecules.
Work of the groups of Tjian and Roeder (44 -48) has indicated that dTAF II 110 makes additional contacts with several other TAFs of the TFIID complex. While most TAF-TAF interactions have not been mapped to particular domains of dTAF II 110, Yokimori et al. (48) demonstrated that dTAF II 30␣ binds to sequences in the carboxyl-terminal part of dTAF II 110 (aa 572-921). In contrast to hPR, dTAF II 30␣ binding was absolutely dependent on the very carboxyl-terminal end of dTAF II 110. Therefore, the sequence requirements for binding of hPR and dTAF II 30␣ to dTAF II 110 are clearly different.
Perhaps the most surprising result of this study is that an hPR construct (MH 6 -C) containing only 18 amino acids of AF-1 (aa 538 -555) and the DBD (aa 556 -639) is sufficient for interaction with dTAF II 110. Since two overlapping mutants (MH 6 -AB⌬core, MH 6 -B⌬core) containing these 18 amino acids of AF-1 are not sufficient for binding to dTAF II 110, the interaction motif is at least partly contained within the DBD of hPR. Another example for an interaction between a DBD and a TAF was recently described by Chiang and Roeder (18), who demonstrated that the DBD of Sp1 binds to a 55-kDa subunit of human TFIID in vitro. Our results thus support the concept that TAFs can interact with multiple distinct domains of transactivators (18).
By cotransfection and in vitro transcription experiments, we were able to demonstrate that the DBD of hPR, which has not been tested for transactivation potential in the studies defining AF-1 and AF-2 of hPR (24,25,38), is able to mediate transcriptional activation through PREs located in front of a minimal promoter, albeit with reduced efficiency compared to wild type hPR (Fig. 7). The presence of a previously unidentified activation function within the hPR domain mediating the interaction with dTAF II 110 in vitro clearly supports our notion that this interaction reflects an important step in the mechanism of transactivation by hPR.
Interestingly, mutational analysis of the rat and human glucocorticoid receptors (49,50) and the Xenopus estrogen receptor (51) has also revealed the presence of a transactivation function within the respective DBDs. While activation by the glucocorticoid receptor DBDs pointed to a role of a basic region immediately following the second zinc finger (49,50), activation by the DBD of the estrogen receptor may be mediated by a short acidic tract at the carboxyl terminus of the DBD (51). Thus, the presence of an activation function located in the DBD might be a more general feature of steroid receptors.
Analogous to the studies that identified interactions of Sp1 and CREB with dTAF II 110 (11,12), we have investigated interactions between a mammalian activator and a Drosophila TAF, i.e. between proteins from evolutionary widely separated organisms. Since the cDNA encoding the human homologue of dTAF II 110 is not available yet, we have not been able to prove that it also interacts with hPR. However, the part of dTAF II 110 sufficient for binding to hPR includes regions with extensive sequence similarity to the human homologue hTAF II 130. 3 Our results thus raise the possibility that this part of TAF II 110 contains a conserved motif mediating interactions with the highly conserved DBDs of members of the steroid receptor superfamily.
Ing et al. (52) have reported that hPR and a truncated chicken PR synthesized in reticulocyte lysate and E. coli, respectively, interact specifically with the basal transcription factor TFIIB. In contrast, using baculovirus-expressed hPR that is transcriptionally active (Fig. 7B), we have been unable to detect association of hPR and TFIIB (Fig. 1B). It remains to be seen whether the different sources of the receptors or experimental differences may account for the discrepancy between the two sets of data. Interestingly, the region required for interaction with TFIIB has been mapped to a 168-amino acid fragment of chicken PR, which includes the highly conserved DBD (52). Since we have shown that the corresponding region of hPR mediates interaction with dTAF II 110, it is feasible that the DBDs of PRs are not only involved in the formation of receptor dimers (21,25), in the recognition of the PREs (21), and in interactions with a TAF, but also in protein-protein interactions with TFIIB (52). Since the DBD of hPR accounts only for a small part of the overall transactivation potential of 3 R. Tjian and N. Tanese, personal communication.  Control transfections were performed using the pSGN expression vector lacking a cDNA insert. CAT activities in untreated (open bars) and progesterone-treated cells (filled bars) were normalized to the values obtained with the empty expression vector. Similar results were obtained in CV1 cells. B, nuclear extracts from rat liver were incubated without recombinant protein or the indicated amounts of E. coli mock extract, MH 6 -hPR(DBD) and MH 6 -hPR0. In vitro transcription reactions were performed as described under "Materials and Methods." Appropriate autoradiographs were scanned with a laser densitometer, and signals representing correctly initiated PRE 2 TATA transcripts were normalized to the internal control (TATA-G400). hPR (Fig. 7), it is likely that hPR contacts additional, yet unidentified components of the initiation complex, possibly through AF-1 and AF-2. Multiple interactions with members of the transcription machinery have been proposed for a growing number of transcriptional activators, including VP16 and the glucocorticoid receptor (13,53). Such a multiplicity of proteinprotein contacts may enable the hPR to affect subsequent ratelimiting steps in the process of preinitiation complex formation and contribute to the highly synergistic activation observed with multiple PRs bound to closely adjacent PREs.