The Intronless and TATA-less HumanTAF II 55 Gene Contains a Functional Initiator and a Downstream Promoter Element*

Human TAFII55 (hTAFII55) is a component of the multisubunit general transcription factor TFIID and has been shown to mediate the functions of many transcriptional activators via direct protein-protein interactions. To uncover the regulatory properties of the general transcription machinery, we have isolated thehTAF II 55 gene and dissected the regulatory elements and the core promoter responsible forhTAF II 55 gene expression. Surprisingly, the hTAF II 55 gene has a single uninterrupted open reading frame and is the only intronless general transcription factor identified so far. Its expression is driven by a TATA-less promoter that contains a functional initiator and a downstream promoter element, as illustrated by both transfection assays and mutational analyses. Moreover, this core promoter can mediate the activity of a transcriptional activator that is artificially recruited to the promoter in a heterologous context. Interestingly, in the promoter-proximal region there are multiple Sp1-binding sites juxtaposed to a single AP2-binding site, indicating that Sp1 and AP2 may regulate the core promoter activity of thehTAF II 55 gene. These findings indicate that a combinatorial regulation of a general transcription factor-encoding gene can be conferred by both ubiquitous and cell type-specific transcriptional regulators.

Human TAF II 55 (hTAF II 55) is a component of the multisubunit general transcription factor TFIID and has been shown to mediate the functions of many transcriptional activators via direct protein-protein interactions. To uncover the regulatory properties of the general transcription machinery, we have isolated the hTAF II 55 gene and dissected the regulatory elements and the core promoter responsible for hTAF II 55 gene expression. Surprisingly, the hTAF II 55 gene has a single uninterrupted open reading frame and is the only intronless general transcription factor identified so far. Its expression is driven by a TATA-less promoter that contains a functional initiator and a downstream promoter element, as illustrated by both transfection assays and mutational analyses. Moreover, this core promoter can mediate the activity of a transcriptional activator that is artificially recruited to the promoter in a heterologous context. Interestingly, in the promoter-proximal region there are multiple Sp1-binding sites juxtaposed to a single AP2binding site, indicating that Sp1 and AP2 may regulate the core promoter activity of the hTAF II 55 gene. These findings indicate that a combinatorial regulation of a general transcription factor-encoding gene can be conferred by both ubiquitous and cell type-specific transcriptional regulators.
The downstream promoter element (DPE), which is located 28 -34 nucleotides downstream of the transcription start site in many Drosophila TATA-less promoters (9,10,24), has a consensus sequence, (A/G)G(A/T)CGTG, and can be recognized by the dTAF II 60 and dTAF II 40 components of Drosophila TFIID (9,24). This finding suggests that TFIID is likely to be the DPE-binding factor. Interestingly, negative cofactor 2 (NC2 or Dr1-Drap1), initially characterized as a TBP-inhibitory activity on a TATA-containing promoter (25)(26)(27)(28), has recently been shown to facilitate transcription from DPE-driven promoters (29). It seems that TFIID and NC2, two of the DPE-acting factors, may work synergistically through the DPE, although their functional relationship remains to be elucidated. Another upstream core promoter element, (G/C)(G/C)(G/A)CGCC, was identified through binding site selection as a GC-rich sequence recognized by TFIIB (30). This TFIIB recognition element (BRE) is located immediately upstream of the TATA box and can be used to modulate preinitiation complex assembly in eukaryotic cells (30) as well as in Archaea (31). Analysis of the promoter database reveals that 57% of the Drosophila core promoters do not contain a TATA box, and the DPE occurs in ϳ40% of the Drosophila promoters (10). Although such statistical data are not yet available for the human genome, it appears that the promoters of human housekeeping genes, oncogenes, growth factors, and transcription factors often lack a TATA box (32). In addition, many natural promoters contain distinct combinations of core promoter elements whose differential utilization plays an important role in regulating gene expression in a spatial, temporal, or lineage-specific manner (13,33,34).
Human TAF II 55 (hTAF II 55) was first identified as an RNA polymerase II-specific TBP-associated factor (TAF II ) in TFIID (35,36) and, like many other TAF II s, was also detected in the TBP-free-TAF II -containing complex (37). However, TAF II 55 is not present in some other TAF II -containing complexes, such as human PCAF (38) and yeast SAGA complexes (39), suggesting that TAF II 55 has unique properties distinct from its role as a structural component of TFIID and of TBP-free-TAF II -containing complex. This idea is further substantiated by the finding that TAF II 55 can interact with many transcription factors, including Sp1, YY1, USF, CTF, adenovirus E1A, and HIV-1 Tat (35), and can also mediate vitamin D 3 and thyroid hormone receptor activation in a ligand-independent manner (40), consistent with a coactivator role of TAF II 55 in transcriptional regulation. Moreover, TAF II 55 may be implicated in mRNA 3Ј end processing, as it shows strong affinity toward the human cleavage-polyadenylation specificity factor (41).
TAF II 55 homologues have also been identified in several organisms. The mouse homologue, mTAF II 55, is 95% identical to its human counterpart (42), and the Saccharomyces cerevisiae homologue, yTaf67, is essential for cellular viability 2 (43). Recently, the Schizosaccharomyces pombe homologue of yTaf67, Ptr6p (poly(A) ϩ RNA transport), was shown to be involved in nucleocytoplasmic transport of mRNAs during a genetic screen for mutants that accumulate mRNAs in the nucleus (44). Moreover, proteins that share high sequence homology with hTAF II 55 have also been identified in Caenorhabditis elegans (GenBank TM accession number Z67755) and Drosophila melanogaster (GenBank TM accession number AF017096). The chromosomal location of the hTAF II 55 gene has been mapped to 5q31, where chromosomal mutations have been associated with stomach adenocarcinoma (45), suggesting that hTAF II 55 or other genes localized in this region may act as an oncogene.
Interestingly, Northern blot analysis showed that hTAF II 55 is differentially expressed in various human tissues. 3 In addition, we observed that in a HeLa-derived cell line that conditionally expresses FLAG-tagged hTAF II 55, the overall level of the induced tagged protein and the endogenous untagged hTAF II 55 protein remains constant (46). This indicates a tight regulation over hTAF II 55 expression in vivo. In order to understand the regulation of hTAF II 55 gene expression and to gain further insight into the regulatory pathways of general transcription factor-encoding genes, we dissected the cis-acting elements and trans-acting factors that regulate the expression of the hTAF II 55 gene. Our studies indicate that hTAF II 55 gene expression is combinatorially regulated by both ubiquitous and cell type-specific transcription factors. Moreover, we have characterized the core promoter elements of the hTAF II 55 gene, which surprisingly contains a single uninterrupted open reading frame whose expression is driven by a TATA-deficient promoter with a functional initiator and a DPE. Collectively, these findings uncover unusual features of hTAF II 55 gene structure and regulatory properties that are significantly different from other general transcription factor-encoding genes.

EXPERIMENTAL PROCEDURES
Isolation of Human TAF II 55 Genomic Clones-A human genomic library, derived from the HT1080 human fibrosarcoma cell line and cloned in the -DASH II vector (Stratagene), was screened with a 32 P-labeled DNA fragment spanning the first 474 nucleotides (cut between the HpaI and EcoRI sites) of the hTAF II 55 cDNA (35). From ϳ1 ϫ 10 6 plaque-forming units, 12 positive clones were isolated. The inserts were individually cloned into the NotI site of pBS-SK (ϩ) (Stratagene). A clone, pBS/3Ј-8, which contains an insert of ϳ17 kb, including regions 5Ј and 3Ј of hTAF II 55, was manually sequenced (GenBank TM accession number AF349038).
The plasmids pGL2-TAF55(Ϫ748/ϩ87) and pGL2-TAF55(Ϫ281/ϩ87) were created by first cleaving pGL2-TAF55(Ϫ1372/ϩ87) with ScaI or XbaI, filling in the XbaI-digested end with Klenow enzyme, and releasing the inserts with BglII. The promoter-containing fragments were then cloned into pGL2-Basic between the BglII site and the Klenowfilled-in XhoI site to generate pGL2-TAF55(Ϫ748/ϩ87) and pGL2-TAF55(Ϫ281/ϩ87), respectively. The plasmid pGL2-TAF55(Ϫ161/ϩ87) was generated by cloning a PCR fragment, amplified with an upstream primer spanning Ϫ161 to Ϫ144 and the same downstream BglII sitecontaining primer ending at ϩ87, between the BglII site and the Klenow-filled-in XhoI site of pGL2-Basic. Similarly, the plasmid pGL2-TAF55(Ϫ1372/Ϫ140) was made by inserting a PCR fragment, amplified with the same upstream KpnI site-containing primer ending at Ϫ1372 and a downstream primer spanning Ϫ157 to Ϫ140, between the KpnI site and the Klenow-filled-in XhoI site of pGL2-Basic.
Transient Transfection and Reporter Gene Analysis-C-33A cells, which were derived from human cervical carcinoma, were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum in humidified 5% CO 2 incubator at 37°C. Transient transfection was carried out in C-33A cells with 4 g of each reporter plasmid, either alone or in conjunction with varying amounts of the Gal4-VP16-expressing plasmid (pSGVP) supplemented with the cloning vector (pSG424) to a total of 1 g, using the calcium phosphate precipitation method as described (53). The transfected cells, after rinsing twice with 1ϫ PBS, were collected 24 h post-transfection by a rubber policeman and resuspended in 100 l of T250E5 buffer (250 mM Tris-HCl, pH 7.6, and 5 mM EDTA). Cell lysates were then prepared by three cycles of freezing and thawing in liquid nitrogen and a 37°C water bath. Following centrifugation at 4°C for 10 min, 2 l of the supernatant was mixed with 350 l of luciferase buffer (25 mM HEPES, pH 7.8, 5 mM ATP, 15 mM MgSO 4 ) with luciferase assays conducted by automatically injecting 100 l of 0.2 mM luciferin (Analytical Luminescence Laboratory) into the samples and measuring the luminescence for 12 s after an initial 2-s delay, using a Monolight 2010 luminometer (Analytical Luminescence Laboratory). Transfection and reporter gene assays were performed independently at least four times, each in duplicate.
In Vitro Transcription and Primer Extension-In vitro transcription was performed with HeLa nuclear extracts and analyzed by primer extension as described (50). The Luc-5 primer (5Ј CTCTTCATAGCCT-TATGCAG 3Ј) and the Luc-1 primer (5Ј TCTTTATGTTTTTGGCGTCT 3Ј) that anneal to nucleotides 151-170 and 81-100, respectively, of pGL2-Basic were used for examining products derived from hTAF II 55 promoter-containing constructs, whereas a chloramphenicol acetyltransferase primer (5Ј CAACGGTGGTATATCCAGTG 3Ј) that anneals to nucleotides 4936 -4953 of pSV2CAT (54) was used for determining the product derived from pHIVϩ58. All the primer extension products were analyzed on an 8 M urea, 5% Long Ranger (FMC) polyacrylamide gel together with the dideoxynucleotide sequencing products generated with the phosphorylated forms of the corresponding primers.
RNase Protection Assay-Total cellular RNA was prepared from eight 100-mm plates of 80% confluent C-33A cells by guanidinium thiocyanate/phenol extraction method using 8 ml of Trizol reagent (Life Technologies, Inc.) according to the manufacturer's instructions. Poly(A) ϩ RNA was isolated by first passing heat-treated total cellular RNA, after mixing with an equal volume of 2ϫ loading buffer, through a 1-ml oligo(dT)-cellulose (Amersham Pharmacia Biotech) column, which was pre-equilibrated with 1ϫ loading buffer (20 mM Tris-HCl, pH 7.6, 0.5 M LiCl, 1 mM EDTA, and 0.1% SDS). The flow-through fraction was collected, denatured at 65°C for 5 min, chilled on ice, and loaded again onto the column. This process was repeated for two additional times. The column was then washed with 6 -8 column volumes of 1ϫ loading buffer. Poly(A) ϩ RNA was eluted with 1 column volume of elution buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 0.05% SDS) for a total of three times, precipitated with ethanol, and finally resuspended in diethyl pyrocarbonate-treated water.
An antisense riboprobe, corresponding to ϩ87 to Ϫ128 of the hTAF II  The reaction was conducted at 37°C for 60 min. The riboprobe was then separated on a 4% polyacrylamide-8 M urea gel, eluted from the gel slice in elution buffer (0.5 M ammonium acetate, 10 mM magnesium acetate, 0.1% SDS, and 1 mM EDTA), extracted with phenol/chloroform, precipitated with ethanol, and finally dissolved in 50 l of 1ϫ hybridization buffer (40 mM PIPES, pH 6.7, 0.4 M NaCl, and 1 mM EDTA). RNase protection assay was carried out as described previously (55) with minor modifications. Briefly, ϳ5 ϫ 10 5 cpm of the in vitro synthesized riboprobe was mixed with 3 g of poly(A) ϩ RNA in a 30-l reaction mixture containing 80% formamide in a final 1ϫ hybridization buffer, overlaid with mineral oil, heated at 90°C for 10 min, and hybridized at 58°C overnight. The hybridization reaction was then quenched on dry ice and incubated with 350 l of RNase solution containing 14 g of RNase A (Sigma), 50 units of RNase T1 (Amersham Pharmacia Biotech), 0.3 M NaCl, 10 mM Tris-HCl, pH 7.5, and 5 mM EDTA at 30°C for 60 min. The ribonucleases were degraded by adding 50 g of proteinase K (U. S. Biochemical Corp.) and 5 l of 10% SDS and incubated for another 15 min at 37°C. The protected fragments were purified by phenol/chloroform extraction, precipitated twice with ethanol, and finally analyzed on a 5% polyacrylamide-urea gel with a DNA sequencing ladder loaded in parallel as size markers. The migration differences between the protected RNA fragments and the DNA size markers were adjusted using undigested riboprobe as standard.

A 1.4-kb Genomic Fragment Preceding the 5Ј End of the
Human TAF II 55 cDNA Sequence Has Intrinsic Promoter Activity-To understand the regulatory properties of general transcription factor-encoding genes, we isolated a dozen genomic clones from an HT1080 human fibrosarcoma genomic library using probes derived from hTAF II 55 cDNA (35). One of the isolated clones, 3Ј-8, containing the entire open reading frame and flanking regions was completely sequenced (17,042 bp, GenBank TM accession number AF349038). The hTAF II 55 gene, which encodes a component of the eukaryotic core promoterbinding factor TFIID, encompasses the complete cDNA sequence of hTAF II 55, suggesting that it is an intronless gene (Fig. 1A). This finding is surprising, given the fact that the human general transcription factor-encoding genes so far identified, including TFIIA (␣/␤ and ␥), TFIIB, TFIIE␣, TFIIE␤, the RAP30 and RAP74 subunits of TFIIF, components (p89, p80, p62, p52, p44, p34, CDK7, cyclin H, and MAT1) of TFIIH, TBP, and other TAF II s in TFIID, all have introns (data not shown). The possibility that our hTAF II 55 genomic DNA was derived from retrotransposition of the hTAF II 55 cDNA was excluded for the following reasons. First, a poly(A) tail sequence found at the 3Ј end of the hTAF II 55 cDNA (35) is absent in all of our genomic clones. The hTAF II 55 sequences identified in the genomic clones and the cDNA diverge at the 3Ј cleavage site where poly (A) addition occurs (data not shown), indicating that reverse transcription and retroviral insertion are unlikely to be involved in generating the genomic copy. Second, all of our independent clones that extended beyond the 3Ј end of the hTAF II 55 cDNA had identical sequences, and all of them lack a poly(A) tail. Third, a portion of the BAC clone 249h5 (Gen-Bank TM accession number AC005618), derived from human chromosome 5, has a nearly identical nucleotide sequence with that of our 3Ј-8 genomic clone. Fourth, the human protocadherin-␥ A1 gene sequence found at the 5Ј end of the 3Ј-8 clone is also present in human chromosome 5, indicating the authenticity of our isolated hTAF II 55 genomic sequence. Finally, the entire sequence of our 3Ј-8 clone is also found in the just-  (35,36) and its corresponding mRNA, relative to the genomic clone, are indicated by a thin line and a thick line, respectively. The 1459-bp genomic fragment encompassing the 5Ј end of the hTAF II 55 cDNA upstream of a luciferase reporter gene as in pGL2-TAF55(Ϫ1372/ϩ87) is also depicted. B, the 1.4-kb genomic fragment upstream of the 5Ј end of the human TAF II 55 cDNA has intrinsic promoter activity. Human cervical carcinoma C-33A cells were transiently transfected with reporter plasmids pGL2-TAF55(Ϫ1371/ ϩ87), pGL-HIVϩ80, pGL7072-161, or pGL2-Control, which contains the hTAF II 55 promoter from Ϫ1372 to ϩ87 (Ϫ1372/ϩ87), the HIV-1 promoter from Ϫ167 to ϩ80 (Ϫ167/ϩ80), the human papillomavirus type 11 (HPV-11) promoter spanning 7072-7933/1-161 (7072/161), or the SV40 promoter/enhancer (Pro/Enh), respectively. The pGL2-Basic plasmid (vector), used for constructions of the above-mentioned reporter plasmids, was also included as control. The luciferase activity was determined as described under "Experimental Procedures" and normalized to that of the hTAF II 55 promoter construct. deposited human genome databases (56,57). Taken together, the absence of a poly(A) tail in our genomic clones and the co-localization of the entire genomic sequence in a single chromosomal locus exclude the possibility that our genomic clones are artifacts and further confirm that the hTAF II 55 gene is indeed devoid of introns, an unusual feature distinct from all the other general transcription factor-encoding genes so far identified.
To identify a functional promoter in the isolated hTAF II 55 gene, we cloned a 1.4-kb genomic DNA fragment that extends 1436 bp upstream and 23 bp downstream of the 5Ј end of the hTAF II 55 cDNA (35) into pGL2-Basic (Fig. 1A). The promoter activity of the resulting construct, pGL2-TAF55(Ϫ1372/ϩ87), was examined by luciferase assays in a human cervical carcinoma-derived C-33A cell line following transient transfection.
As shown in Fig. 1B, the 1.4-kb genomic fragment of hTAF II 55 has promoter activity that is stronger than those exhibited by HIV-1, human papillomavirus type 11 (HPV-11), and SV40.
Mapping the Transcription Start Site of the hTAF II 55 Promoter-In order to locate the transcription start site of the hTAF II 55 gene, we first performed in vitro transcription with HeLa nuclear extracts, using pGL2-TAF55(Ϫ1372/ϩ87). The in vitro synthesized transcripts were then detected by primer extension analysis ( Fig. 2A). To minimize artifacts caused by spurious primer annealing, we used two primers, Luc-1 and Luc-5, that anneal to different positions of the transcript and are expected to generate ϳ150and 210-nucleotide products, respectively. When either primer was used, the transcription start site was mapped to the same position in the genomic sequence (nucleotide 12,849), which was designated ϩ1 (Fig.  2A, lanes 1 and 3, indicated by an arrow). Several signals of less intensity corresponding to ϩ3 to ϩ6 positions were also detected. The presence of multiple minor transcription start sites in a TATA-less promoter is not uncommon (see Refs. 9, 24, and 58 -60; and see below). A control template with the TATAcontaining HIV-1 promoter was mapped to the same start site as determined previously (Fig. 2A, lane 5; Ref. 49). The transcripts derived from the hTAF II 55 and HIV-1 promoters are RNA polymerase II-specific, since the addition of a low concentration (2 g/ml) of ␣-amanitin, which inhibits the activity of RNA polymerase II, completely abolished the specific signals ( Fig. 2A, compare lanes 1 and 2, 3 and 4, and 5 and 6).
A transcription signal detected in the Luc-5 experiment ( Fig In vitro transcription was conducted with HeLa nuclear extracts using the hTAF II 55 promoter-containing construct pGL2-TAF55(Ϫ1372/ϩ87) or the HIV-1 promoter-containing construct pHIVϩ58, in the absence (Ϫ) or presence (ϩ) of 2 g/ml of ␣-amanitin. Two primers, Luc-1 and Luc-5, whose relative positions and expected sizes of primer extension products are indicated at the bottom, were used to determine the start site of the hTAF II 55 gene, whereas a chloramphenicol acetyltransferase (CAT) primer that anneals to the chloramphenicol acetyltransferase reporter gene was used for mapping the start site of the HIV-1 promoter. DNA sequencing ladders, prepared from the phosphorylated forms of the corresponding primers and DNA templates as employed for in vitro transcription, were included for the assignment of the transcription start sites (indicated by arrows). The DNA sequences surrounding the transcription start sites are shown on the left of each panel with a bent arrow pointing to the major start site at ϩ1 and solid squares indicating relative intensities of the transcription signals. Two reproducible transcription signals, mapped to an upstream (indicated by an arrowhead) or downstream (indicated by an asterisk) location of the hTAF II 55 cDNA, are marked on the right of the panels. B, RNase protection analysis of in vivo hTAF II 55 transcripts. RNase protection assays were performed by first hybridizing in vitro synthesized antisense riboprobe of 285 nt spanning Ϫ128 to ϩ87 with endogenous poly(A) ϩ RNA isolated from C-33A cells or with tRNA. RNase A and RNase T1 were then added to digest the single-stranded region. The protected fragments, along with a DNA size marker (A, C, G, and T) and the original riboprobe (Ϫ) used to adjust the migration difference between DNA and RNA, were then analyzed on a 5% polyacrylamide, 8 M urea gel and visualized after exposure to an x-ray film. The positions of the major protected fragment (87 nt) and the riboprobe are indicated, respectively, by arrows. sion, since it lies within the isolated cDNA region (36). Another transcription signal detected in the Luc-1 experiment (Fig. 2A, lane 1, shown with an arrowhead) is located at Ϫ57, surrounded by GC-rich sequences. It might represent an alternative transcription start site or a spurious transcript caused by nonspecific initiation of RNA polymerase II in vitro. To distinguish between these two possibilities, we isolated endogenous poly(A) ϩ RNA from C-33A cells and determined the transcription start site using RNase protection assay with an antisense RNA probe that spans nucleotides from ϩ87 to Ϫ128 relative to the transcription start site (Fig. 2B). A correct initiation at ϩ1 would give rise to a protected fragment of 87 nucleotides. In contrast, were transcription initiated from Ϫ57, a 144-nt protected fragment would be detected. As shown in Fig. 2B, only an 87-nt protected fragment, corresponding to the start site mapped in vitro, was observed when the 32 P-labeled riboprobe was hybridized with poly(A) ϩ RNA but not with tRNA (lanes 1  and 2). The absence of a 144-nt protected fragment suggests that the start site detected at Ϫ57 is an artifact caused by nonspecific initiation of RNA polymerase II in vitro. Therefore, we concluded that nucleotide 12,849 in the 3Ј-8 genomic clone is the major transcription start site of the hTAF II 55 promoter both in vivo and in vitro.
Transcription Factors Potentially Regulate hTAF II 55 Promoter Activity-A search for transcription factors potentially regulating hTAF II 55 gene expression was performed using the MatInspector program. We found putative binding sites for STAT-1, MEF2, E2F, Sp1, AP2, AREB6, and E47 in the promoter-proximal region (Fig. 3A). Obviously, no TATA box is located between Ϫ25 and Ϫ30, but, instead, there are consensus Inr and DPE sequences surrounding the transcription start site and spanning ϩ29 to ϩ35, respectively. This inspection reveals that an intrinsic TATA-less promoter is used for hTAF II 55 gene expression, which is likely regulated by both ubiquitous and cell type-specific transcription factors.
Sequences for Sp1, AP2, AREB6, and E47 Binding Are Important for hTAF II 55 Gene Expression-To define the transcription factor-binding sites that were important for hTAF II 55 gene expression, we made a series of 5Ј deletion constructs and tested their promoter activity following transfection into C-33A cells. As shown in Fig. 3A, deletions progressing to Ϫ99 that removed the STAT-1, MEF2, and E2F sites showed no significant reduction in promoter activity in C-33A cells. Further deletion of the region from Ϫ99 to Ϫ71, which contains no known transcription factor-binding sites, resulted in ϳ2-fold decrease in promoter activity. Interestingly, deletion up to Ϫ55, which removes a putative Sp1-binding site centering on Ϫ60, caused another 2-3-fold reduction. An additional deletion to Ϫ26, which eliminates an overlapping Sp1-and AP2-binding site at Ϫ50, resulted in an extra 8-fold reduction. In contrast, an upstream fragment spanning Ϫ1372 and Ϫ140 showed no promoter activity, further confirming our results of start site mapping.
A similar conclusion was also obtained by in vitro transcription and primer extension assays performed with HeLa nuclear extracts using similar hTAF II 55 deletion constructs. As shown in Fig. 3B, whereas a series of 5Ј deletions up to Ϫ128 did not markedly affect promoter functioning, the Ϫ71/ϩ87 construct reduced ϳ50% of the promoter activity. Further deletions to Ϫ55 and Ϫ26 significantly decreased the signal intensity. However, the transcription start site initiating at ϩ1 was still detectable after longer exposure. This analysis suggests that FIG. 3. Promoter sequences containing Sp1-and AP2-binding sites are important for hTAF II 55 promoter activity. A, reporter gene assays performed in C-33A cells with hTAF II 55 promoter constructs that sequentially remove potential transcription factor-binding sites. Transient transfection and reporter gene assays were performed as described under "Experimental Procedures" using plasmids containing the hTAF II 55 promoter sequences with the indicated boundaries. The pGL2-Basic plasmid containing no insert was also used for transfection as control. Luciferase activity was normalized to that of the full-length promoter construct, pGL2-TAF55(Ϫ1372/ϩ87), and presented in the bar graph with error bars showing standard deviation. B, in vitro transcription and primer extension analysis of hTAF II 55 promoter deletion constructs with HeLa nuclear extracts. In vitro transcription was performed in HeLa nuclear extracts with plasmids containing the hTAF II 55 promoter sequences as indicated. The in vitro synthesized transcripts were then mapped by primer extension using Luc-1 primer and analyzed on a 5% polyacrylamideurea gel. The products with correct transcription start sites at ϩ1 are indicated by an arrow. DNA sequencing ladders, prepared from pGL2-TAF55(Ϫ1372/ϩ87) using the 5Ј-phosphorylated Luc-1 primer and [␣-35 S]dATP, were included to determine the position of the correctly initiated transcripts. Schematic diagrams of the promoter deletion constructs used for both experiments are drawn on the left, with potential transcription factor-binding sites marked in boxes.
both Sp1 and AP2 sites are important for hTAF II 55 gene expression. Moreover, both in vitro and in vivo assays indicated that the Ϫ26/ϩ87 construct, although it shows much weaker promoter activity compared with that from the 1.4-kb genomic fragment, could still direct reporter gene expression, suggesting that the critical core promoter elements essential for hTAF II 55 promoter activity are retained in this short region (see below).
To examine whether the downstream region containing putative AREB6-and E47-binding sites are also critical for hTAF II 55 gene expression, we created several 3Ј deletion constructs removing the sequence between ϩ36 and ϩ87 and tested their promoter activity in C-33A cells by transfection assays. As shown in Fig. 4A, promoter constructs deleted to ϩ36 reduced reporter gene activity ϳ3-fold. This result reveals that the sequences downstream of the transcription start site, including the AREB6-and E47-binding sites, contribute to hTAF II 55 promoter activity. Nevertheless, for core promoter activity, the region between ϩ36 and ϩ87 seems dispensable. The Ϫ55/ϩ36 construct still maintained promoter activity sufficient to drive reporter gene expression.
From the deletion analysis, it appears that Sp1 and AP2 likely play an important role in optimizing hTAF II 55 promoter activity, which could be conferred by a small DNA fragment spanning Ϫ26 to ϩ36. To test this hypothesis, we created the Ϫ26/ϩ36 promoter construct and compared its promoter activity with several 5Ј deletion constructs all ending at ϩ36 as well as with new constructs containing nucleotide substitutions in the DNA-binding sites for Sp1 and AP2. As shown in Fig. 4B, nucleotide substitutions at the Ϫ60 Sp1-binding site reduced promoter activity 2-3-fold (compare 2nd and 3rd constructs), consistent with the result from the 5Ј deletion constructs (see Fig. 3A, compare Ϫ71/ϩ87 and Ϫ55/ϩ87 constructs). However, mutations introduced at the AP2-binding site showed only 10 -20% reduction of reporter activity (Fig. 4B, compare  2nd and 4th constructs and 3rd and 5th constructs). Interestingly, mutations at the Ϫ20 Sp1-binding site showed the same activity as that of the wild-type construct (Fig. 4B, compare 2nd  and 6th constructs). This finding indicates that different promoter-proximal Sp1-binding sites contribute unequally to hTAF II 55 promoter activity. As expected, the Ϫ26/ϩ36 construct still retains promoter activity (Fig. 4B). A similar result was also obtained with in vitro transcription assays performed with HeLa nuclear extracts (data not shown).
The hTAF II 55 Core Promoter Is TATA-less with Functional Inr and DPE Sequences-The finding that a DNA fragment spanning Ϫ26 to ϩ36 still retained hTAF II 55 promoter activity and that mutations introduced at the Ϫ20 Sp1-binding site had no effect on reporter activity (Fig. 4B) suggested that the Inr and DPE motifs present in this region were likely to be the functional modules driving hTAF II 55 gene expression. To test this, we introduced mutations in the Inr and DPE, either individually or in combination, and tested the activity of the core promoter constructs using transfection assays in C-33A cells. As shown in Fig. 5A, mutations in the Inr and the DPE reduced promoter activity ϳ33and 5-fold, respectively, whereas double mutations essentially abolished the promoter function.
To verify that the Inr and the DPE identified in the hTAF II 55 promoter can function as independent promoter modules, we introduced five Gal4-binding sites into the wild-type and mu- FIG. 4. The hTAF II 55 promoter is regulated by proteins targeting the AREB6-, E47-, Sp1-, and AP2-binding sites in C-33A cells. A, reporter gene assays performed in C-33A cells with hTAF II 55 promoter constructs with or without the AREB6-and E47-binding sites. Transient transfection and reporter gene assays were performed as described under "Experimental Procedures" using plasmids containing the hTAF II 55 promoter sequences with the indicated boundaries. The pGL2-Basic plasmid containing no insert was also used for transfection as control. Luciferase activity was normalized to that of the full-length promoter construct, pGL2-TAF55(Ϫ1372/ ϩ87), and presented in the bar graph with error bars showing standard deviation. B, nucleotide substitutions in the Sp1-and AP2-binding sites reduce hTAF II 55 promoter activity in C-33A cells. Transient transfection and reporter gene assays were performed with plasmids containing the hTAF II 55 promoter sequences with the indicated boundaries. Asterisks indicate mutations introduced at specific protein-binding motifs in the plasmids. The nucleotides changed in each motifs are denoted at the bottom.
tated hTAF II 55 core promoter constructs, and we tested promoter activity by cotransfection with a Gal4-VP16-expression plasmid. As shown in Fig. 5B, expression of Gal4-VP16 significantly enhanced wild-type (WT) hTAF II 55 promoter activity in a dose-dependent manner. In contrast, Gal4-VP16 had little, if any, effect on constructs containing Inr mutation (Inr*) or Inr and DPE double mutations (Inr*DPE*). The heterologous promoter with the DPE mutation (DPE*) showed a slight response to Gal4-VP16. This result demonstrates that the Inr and the DPE derived from the hTAF II 55 promoter are indeed core promoter elements that can mediate the activity of a transcrip-tional activator artificially recruited to the promoter in a heterologous context. DISCUSSION In this report, we describe the detailed characterization of regulatory elements and core promoter critical for the expression of the human TAF II 55 gene, which encodes a component of the general transcription factor TFIID. Sequencing of our isolated hTAF II 55 genomic clones and mapping of the transcription start site reveal that the hTAF II 55 gene is intronless, a feature distinct from the other general transcription factorencoding genes so far identified (56,57,(61)(62)(63). Furthermore, expression of hTAF II 55 is driven by a TATA-less promoter with a functional Inr and the DPE active in both homologous and heterologous promoter contexts.
Intronless Genes-In higher eukaryotes, most genes contain introns. Compared with 96% intronless genes in S. cerevisiae, there are 17% intronless genes in D. melanogaster and merely 6% of genes in mammals without introns (64). One family of intronless genes encodes histones, which are comparatively small, abundantly expressed and highly conserved in sequence (65). Another family encodes G-protein-coupled receptors (66).
Since intronless genes such as those encoding hsp70 (67), c-jun (68), and interferon-␣ (69) do not require post-transcriptional splicing, they may be expressed more efficiently and are believed to be involved in immediate response to extracellular signals. On the other hand, many viruses that undergo reverse transcription during the replication cycle have evolved special mechanisms to facilitate specifically export of intronless gene products to the cytoplasm and inhibit the splicing process (70,71). Considering the hTAF II 55 gene has multiple STAT-and E2F-binding sites in its promoter-proximal region, it seems probable that hTAF II 55 is involved in integrating extracellular signals to the general transcription machinery. This point of view is further supported by the finding that hTAF II 55 can interact with many transcription factors (35) and can also mediate the functions of several nuclear hormone receptors (40).
Core Promoter Elements-One interesting property of the hTAF II 55 promoter sequence is that it has no cognate TATA box or even AT-rich sequences within the promoter-proximal 80 nucleotides upstream of the transcription start site. Instead, it contains a consensus Inr that overlaps the transcription start site and the DPE core sequence (GGACGGA) from ϩ29 to ϩ35. Both the Inr and the DPE are critical for hTAF II 55 core promoter function, as illustrated by both transfection assays (Fig.  5) and in vitro transcription analysis (data not shown). The Inr is clearly protected by proteins present in nuclear extracts (data not shown), consistent with the functional importance of the Inr (Fig. 5). Although mutations at the DPE did not completely abolish promoter function, these constructs displayed dramatic decreases in promoter activity. We speculate that the incomplete destruction of the Sp1-binding site at the Ϫ20 region in the Ϫ26/ϩ36 promoter-based constructs might partially compensate for the loss of the DPE function. It is also likely that the Inr and the DPE of the hTAF II 55 gene as well as the Ϫ20 Sp1-binding site are differentially utilized in different cell types. Therefore, the DPE may be more important in some cells than in others. Nevertheless, this study is the first demonstration of a functional DPE in a human promoter following the initial report on hIRF (24).
MED-1 (multiple start site element downstream) in many TATA-less promoters and DCE (downstream core promoter element) in the TATA-containing human ␤-globin promoter are additional examples of downstream elements that function in concert with the Inr and appear to affect TFIID binding (58,72). Cellular proteins, such as TFIID and NC2, have been FIG. 5. Inr and DPE are both important core promoter elements for hTAF II 55 gene expression. A, mutations at the Inr and the DPE reduce hTAF II 55 promoter activity. Transient transfection and reporter gene assays were performed as described under "Experimental Procedures" using plasmids containing either wild-type or mutated nucleotides at the Inr and/or the DPE of the hTAF II 55 promoter fragment spanning Ϫ26 to ϩ36. The pGL2-Basic plasmid (vector) containing no insert was also used for transfection as control. Luciferase activity was normalized to that of the wild-type promoter construct, pGL2-TAF55(Ϫ26/ϩ36), and presented in the bar graph with error bars showing standard deviation. Asterisks and ϫ indicate mutations introduced at specific protein-binding motifs in the plasmids. The nucleotides changed in each motifs are denoted at the bottom. B, the Inr and DPE modules of the hTAF II 55 core promoter can mediate transcriptional activation in a heterologous promoter context. Transient transfection was performed in C-33A cells by cotransfecting different amounts of the Gal4-VP16-expressing plasmid (pSGVP), together with either wild-type (WT) or mutated reporter constructs driven by 5 Gal4binding sites as indicated.
reported to act through these downstream promoter elements. However, it is not clear whether the downstream sequences are essential for promoter activity merely to affect preinitiation complex formation or whether they are also involved in promoter clearance and the formation of a highly processive RNA polymerase II elongation complex (73).
Our present study provides convincing evidence that the Ϫ26/ϩ36 sequence can serve as an independent core promoter module, which can be further activated by a transcriptional activator in the context of a heterologous promoter (Fig. 5B). The hTAF II 55 core promoter could thus be a model to analyze further molecular mechanisms of transcription initiation on TATA-less promoters.
Transcription Factors Binding to the Promoter-proximal Region-Our study also details transcriptional regulation of hTAF II 55 promoter activity by Sp1 and AP2 proteins, a phenomenon commonly observed in mammalian TATA-less promoters (74,75). Sp1 is a well characterized ubiquitous transcription factor whose binding sites are found in numerous promoters that regulate both ubiquitous and tissue-specific genes (60, 76 -78). Mice with homozygous deletions of the Sp1coding gene show severe developmental defects and die early during embryogenesis, suggesting that Sp1 is essential for embryonic development (79). Our study also indicates the importance of Sp1 in regulating TAF II 55 gene expression. First, deletion of the GC-rich Sp1-binding sequences resulted in a significant reduction in hTAF II 55 promoter activity (Fig. 3). Second, point mutations introduced at the Sp1-binding sites resulted in similar decreases in activity (Fig. 4). Third, DNase I footprinting shows direct binding of purified Sp1 to its cognate DNA sequences (data not shown).
In contrast, AP2 is a cell type-specific transcription factor important in retinoid-controlled morphogenesis and differentiation, especially in neural crest-derived cell lineages and epithelial cells (80). AP2 responds to at least two different signal transduction pathways, the phorbol ester/protein kinase C signaling and the cAMP-dependent protein kinase pathway (81). AP2 has a spatially and temporally restricted expression pattern in murine embryos and shows significant expression levels in adult skin and urogenital tissues (80). We found that Sp1 and AP2 proteins, whose binding sites are closely positioned on the hTAF II 55 promoter, could bind simultaneously to the promoter (data not shown). This finding suggests that Sp1 and AP2 can regulate the hTAF II 55 promoter in a combinatorial manner, although they do not appear to function synergistically in C-33A cells (Fig. 4B). We estimate by quantitative Western blotting analysis that C-33A cells have ϳ100 fg of Sp1 and less than 5 fg of endogenous AP2 per cell (data not shown). It is likely that the relatively low level of AP2 proteins in C-33A cells cannot confer significant activator function on hTAF II 55 expression. AP2 may function as a more potent transcription activator in keratinocytes or in the neural crest lineage in which it is expressed in high levels. We will further clarify the role of AP2 in regulating hTAF II 55 gene expression by overexpressing AP2 in C-33A cells or by performing transfection assays in different cell types.
The discovery of many potential transcription factor-binding sites in the hTAF II 55 promoter-proximal region raises several interesting issues. Many TATA-less genes involved in DNA replication and cell cycle control have been reported to contain E2F-and Sp1-binding sites (77). Colocalization studies of cells at different stages of the cell cycle indicate that Sp1 may physically and functionally associate with E2F (77,82). It remains to be investigated whether the E2F proteins also functionally interact with Sp1 on the hTAF II 55 promoter. Intriguingly, the presence of a MEF2-binding site may account for the preferential expression of the TAF II 55 mRNA in skeletal muscle, as revealed in Northern blotting analysis. Although transfection assays and in vitro transcription carried out in human cervical cancer cell lines (HeLa or C-33A) did not reveal the functional importance of E2F-, STAT-, and MEF2-binding sites in the hTAF II 55 promoter, we cannot exclude the possibility that these factors are either limiting in our cells or they require additional cellular factors to support activator function. These interesting possibilities remain to be addressed in the future.