INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Transcription initiation in metazoan genes is directed by the core promoter elements that comprise the TATA box and/or the pyrimidine-rich initiator (Inr) element (36). The heterogeneity in core promoter elements might allow alternate initiation pathways in response to specific regulatory factors in various cellular responses. Thus, it has been shown that specific TATA elements (10, 14, 40, 48) or specific Inr elements (9, 12, 17) are important for certain responses, and substitution of one element for another may in fact result in deregulation and loss of developmental or cell-type specificity (reviewed in reference 31).

The TATA-directed basal (activator-independent) transcription (6, 33) begins with TATA recognition by the TATA-binding protein component of transcription factor TFIID. This step is sufficient to nucleate the assembly of additional general transcription factors and RNA polymerase II into a functional complex (5, 36). However, the corresponding pathways for Inr-directed basal transcription appears to be more complex and less well understood (13, 36). The Inr-mediated basal transcription requires several factors, including TBP-associated factors (TAFs), that are not required for TATA-directed basal transcription (30, 36). Some of these factors were also shown to be required for Inr function in conjunction with TATA elements (24, 30) and in conjunction with known or inferred distal regulatory elements (29, 42); however, the latter studies could not distinguish general TAF requirements for activator functions as observed for TATA-containing promoters (36, 47).

Three distinct models have been proposed for Inr recognition in the absence of a TATA box. One model proposes direct recognition of the Inr by a TAF component and an adjacent downstream promoter element when present (3, 4, 46), followed by stable TFIID binding and subsequent preinitiation complex assembly. However, at least in mammalian promoters, the TAFs do not appear to show any sequence specific interactions at the Inr element (24, 36). Moreover, unlike initial expectations TAF_II150 is clearly not involved in direct Inr recognition (25). A second model proposes that recognition of the Inr by independent Inr binding proteins, followed by secondary interactions with TFIID or associated factors, nucleates assembly of the general transcription factors at the core promoter. Consistent with the latter model, several factors have been shown to bind at an Inr element or sites adjacent to it (43). Yet a third model proposes recognition of Inr by RNA polymerase II in the absence of both TAFs and independent Inr binding proteins (49). These observations could reflect diversity in core promoter Inr elements and corresponding interactions, especially in light of the loose consensus sequence for such elements (21, 36). Given such a diverse set of results, it is clear that identification and characterization of the protein factors directly involved in Inr function not only in vitro but also in vivo are required to clarify the issue.

Our current studies are directed toward resolving this problem, with emphasis on Inr function in vivo in the absence of TATA elements. In our effort to understand the transcriptional regulation of the TATA-less, Inr-containing (TATA Inr⁺) genes in general, we have employed the murine V 5.2 promoter as a model (1). In earlier studies we identified a multifunctional transcription factor, TFII-I, that binds at and functions through pyrimidine-rich Inr elements (22, 37, 38) and was critically required for the transcription of the V promoter in vitro (29, 32). Consistent with its multifunctional potential, recent cDNA cloning and functional expression of recombinant TFII-I demonstrated that TFII-I functions as a basal factor through the Inr element in a TATA- and Inr-containing promoter (39) and as an activator through upstream promoter elements in the absence of a functional Inr (15, 26, 39). These observations suggest that TFII-I is a novel factor that is capable of mediating communication between the regulatory and basal components in a eukaryotic gene (15, 26, 39). More surprisingly, TFII-I has also been cloned as a factor (BAP-135) that is tyrosine phosphorylated and interacts with the B-cell-specific Bruton's tyrosine kinase (50) and as a functional gene that is deleted in William's syndrome (34). In this study we demonstrate that in in vivo (transient-transfection) assays ectopically expressed wild-type TFII-I markedly activates the V promoter in an Inr-dependent fashion. Furthermore, we have biochemically identified and ectopically expressed a DNA binding fragment of TFII-I (p70) that exhibits specific Inr binding properties in vitro but lacks the Inr-dependent transcriptional activation in transient-transfection assays. As expected, given these properties, the p70 mutant behaves in a dominant negative fashion when coexpressed with the wild-type TFII-I. However, the Inr-specific activation functions of p70 can be restored when the activation domain of GAL4 is fused to it. Together, these data clearly demonstrate that TFII-I functions through the Inr element and address for the first time the transcriptional functions of an Inr binding protein on a naturally TATA Inr⁺ promoter in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture. COS7 cells were grown in Dulbecco's modified Eagle medium (DMEM; Cellgro) containing 10% fetal bovine serum (FBS; Sigma), 50 U of penicillin per ml, and 50 µg of streptomycin (GIBCO BRL) per ml at 37°C under 5% CO₂.

Plasmids and antibody. (i) TFII-I derivatives. The eukaryotic expression vector pEBGII-I (pEBG containing TFII-I cDNA) is derived from pEBG (44), in which the human EF-1 promoter drives the expression of protein fused to glutathione S-transferase (GST). To construct pEBGII-I, the open reading frame for TFII-I plus the hexa-histidine tag was isolated from pET11d-II-I (39) by NcoI and ClaI digestion. A BamHI linker was ligated to the NcoI end, and the whole fragment was inserted in frame between the BamHI and ClaI site of pEBG. GST-p70 was made by introducing a stop codon after amino acid (aa) 735 by PCR-based mutagenesis. The region between aa 606 and aa 735 was PCR amplified from pEBGII-I by using 5' GTTGTTAAAAAACCTGAACTAG 3' and 5' GTAAATCGATCTAACCCTCAGGTA 3' as primers. The amplified fragment was gel isolated and restricted with SpeI and ClaI. The digested product was then ligated into pEBGII-I. The bacterial p70 expression construct pET11d-p70 was made as follows. The region between aa 606 and aa 735 was PCR amplified from pET11d-II-I by using 5' GTTGTTAAAAAACCTGAACTAG 3' and 5' AATCCTAGGTATCGATACTAACC 3' as primers. The PCR product was gel purified, digested with SpeI and AvrII and then ligated to pET11d-II-I. To generate the expression plasmid p70-GAL4ADII, the activation domain II (aa 768 to 881) of GAL4 was PCR amplified from pMA242 (28) by using the primers 5' GGACCTGAGGGTTTTAATCAAAGTGGGAAT 3' and 5' ATATGCGGCCGCTATTACTCTTTTTTTGG 3'. The PCR-amplified cDNA was gel purified and digested with AocI and NotI and then ligated with GST-p70 at aa 735. The plasmid containing the C-terminal 280 aa of TFII-I (aa 677 to 957) fused to GAL4 DNA binding domain (aa 1 to 147) was generated as follows. The C280 TFII-I was PCR amplified by using the primers 5' GGGGATCCGTGTGCCATTCCGA 3' and 5' GGGATCTAGAGCTACCACGTGG 3'. The PCR-amplified product was gel purified, digested with BamHI and XbaI, and then ligated in frame with GAL4 DNA binding domain in pSG424 plasmid (a gift from Brent Cochran).

(ii) V reporters. The plasmid containing the V FL-Luc was made by PCR amplification of the region between 473 and 9 of the V 5.2 promoter by using 5' GGGATAAGATCTCCAGGTGGCGCTGTGGAC 3' and 5' GGGTAAGCTTCGGCTCCTCCTTTCTC 3' as primers from PGL2V 5.2-Luc (a gift from Michael Meisterernst). The PCR-amplified product was gel purified, digested with HindIII and BglII, and then ligated with PGL3-Basic (Promega Corp.). The truncated V promoter, V61-Luc, with an intact decamer element (Deca) (2), was constructed by PCR amplifying the specified regions (61 to 9) from VFL-Luc. The PCR primers used were 5' GGGAGATCTAGAACCTGACAT 3' and 5' GGGAGGCTTGAGAAAGTGAGAGT 3'. The initiator mutant containing plasmids VFLMut-Luc and for V61Mut-Luc was generated by PCR-directed mutagenesis. For VFLMut-Luc 5' GGGATAAGATCTCCAGGTGGCGCTGTGGAC 3' and 5' AGGCTTGAGCCCCTGAGCGTCGG 3' were used as primers, and for V61Mut-Luc 5' GGGAGATCTAGAACCTGACAT 3' and 5' AGGCTTGAGCCCCTGAGCGTCGG 3' were used as primers. The amplified PCR fragments were gel purified, digested with HindIII and BglII, and ligated with PGL3-Basic. All clones were confirmed by sequencing.

(iii) Eukaryotic expression and purification of TFII-I. For eukaryotic expression of TFII-I, COS7 cells were transfected with pEBGII-I. Transfections were carried out by the Lipofectamine method according to the manufacturer's protocol (GIBCO BRL). For transfection, 10 µg of expression plasmid (pEBGII-I) was used per 100-mm-diameter plate. At 36 h posttransfection, cells were harvested, washed twice in phosphate-buffered saline (PBS), and lysed by sonication in 2 ml of ice-cold BC500 buffer (20 mM Tris-HCl [pH 7.9], 500 mM KCl, 20% glycerol) containing 0.1% Nonidet P-40 and protease inhibitors (1 mM phenylmethylsulfonyl fluoride; 1% aprotinin, leupeptin, and antipain; and soybean trypsin inhibitor at 1 µg/ml). The lysate was clarified by centrifugation for 30 min at 12,000 rpm at 4°C. The GST-hexahistidine tag TFII-I protein was purified by loading the lysate on a 1-ml Ni²⁺-agarose (Invitrogen) column at 4°C. The column was washed sequentially with 5 column volumes of BC500 (without protease inhibitors and Nonidet P-40) containing 20 and 80 mM imidazole, respectively. Finally, the tagged fusion protein was eluted with 3 column volumes of BC500 containing 200 mM imidazole. The TFII-I-containing peak fractions were pooled, dialyzed against buffer B (20 mM Tris-HCl [pH 7.9], 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 10% glycerol) with 100 mM KCl (B100), and loaded onto a Hitrap SP column (Pharmacia). The Hitrap column was washed sequentially with 5 column volumes of B100 and B200 (buffer B with 200 mM KCl) and eluted with 3 column volumes of B500 (buffer B with 500 mM KCl).

Purification of bacterially overexpressed p70. p70 was overexpressed in the bacterial strain BL21(DE3) pLysS (Stratagene) and purified over Ni²⁺ as described previously (39).

EMSA. The electrophoretic mobility shift assays (EMSAs) shown in Fig. 1, 3, and 5 were performed with an Inr probe (28 to 12) derived from the V promoter (29). The probes in Fig. 2 were derived as follows. A 109-bp DNA fragment (100 to 9) containing wild-type Inr or the Inr mutant was isolated from V100Luc or V100Mut-Luc by digesting the plasmid with HindIII and BglII. The fragment was gel purified and labeled with [-³²P]dATP (3,000 Ci/mmol), [-³²P]dTTP (3,000 Ci/mmol), and Klenow fragment. The EMSA and electrophoresis were done as described previously (29, 32).

Western blot analysis. Either purified TFII-I or extracts were subjected to sodium dodecyl sulfate (SDS)-7.5% polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose by the semidry blotting method in a buffer containing 0.192 M glycine, 0.025 M Tris base, and 20% methanol. The blot was blocked in Tris-buffered saline (10 mM Tris [pH 8.0], 150 mM NaCl) with 6% nonfat dry milk (Carnation). The primary (anti-TFII-I, 1:2,500 dilution) and secondary (1:1,500 dilution) anti-rabbit horseradish peroxidase-linked antibodies were incubated in Tris-buffered saline containing 0.05% Tween 20.

In vitro transcription. The nuclear extract was prepared as described previously (8) and was either mock depleted with preimmune serum or immunodepleted with anti-TFII-I antibody (29). Native (20 ng) or recombinant (60 ng) TFII-I was added as indicated. Processing of the transcription reactions was as described previously (29).

Transient-transfection and luciferase assay. One day before transfection, COS cells at 80 to 90% confluency (in 60-mm plates) were trypsinized for 15 min with 2 ml of trypsin, and the volume was increased to 10 ml with DMEM containing 10% FBS. A 200-µl cell suspension was seeded into each well of a six-well plate. Transfection was done with Lipofectamine as suggested by the manufacturer (GIBCO BRL) with the following modifications. In an Eppendorf tube, various reporter plasmids with or without different TFII-I expression constructs and 1 µg of renilla luciferase plasmid (pRL-TK; Promega Corp) were mixed. The total amount of DNA in each experiment was equalized with empty vector pEBB (44). The final volumes were adjusted to 100 µl with Optimem. In a separate Eppendorf tube, 15 µl of Lipofectamine was mixed with 85 µl of Optimem. The plasmid-containing medium was mixed with the Lipofectamine-containing medium and then incubated at room temperature for 45 min to allow the formation of DNA-lipid complex, during which the cells were washed twice with PBS. At the end of the incubation period, the DNA-lipid complex was diluted with 800 µl of Optimem, and the mixture was added to the cells and incubated overnight in a CO₂ incubator at 37°C. After 12 to 14 h, 2 ml of DMEM containing 22.5% FBS was added to each of the cells and incubated an additional 8 h. The cells were then treated with fresh medium containing 15% serum. At 36 h posttransfection, the cells were washed twice in PBS and then lysed, and the luciferase activities were determined. Luciferase activities in the transfected cells were determined according to the manufacturer's protocol (Dual luciferase assay; Promega Corp). Cells were lysed in 500 µl of passive lysis buffer and centrifuged at 14,000 rpm (4°C) for 2 min, and the supernatant was collected. A 10-µl portion of the supernatant was mixed with 50 µl of luciferase assay reagent II for 10 s, and the luciferase activity was determined. Then, to normalize the transfection efficiency, 50 µl of stop-and-glow buffer was added, and the renilla luciferase activity was determined.

Protease treatment and N-terminal sequencing of TFII-I. Approximately equal amounts of highly purified native and recombinant TFII-I (40 to 60 ng) were digested with 1 CLU of thrombin or V8 protease (7.5 ng) by incubation at room temperature for the indicated time. At the end of the incubation, the digestion was stopped by adding the SDS loading buffer, and the digested product was analyzed on a 7.5% SDS gel and immunoblotted with an anti-TFII-I antibody or silver stained according to the manufacturer's directions (Pharmacia Biotech). For N-terminal sequencing the digested product was transferred onto polyvinylidene difluoride membrane and stained with Coomassie blue. The 70- and 43-kDa bands were cut out and subjected to N-terminal microsequencing at the Tufts University sequencing facility.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Ectopic expression of a transcriptionally competent TFII-I in mammalian cells. To demonstrate the DNA binding and transcriptional activity of recombinant TFII-I, it was ectopically expressed in and purified from COS cells. For this purpose, we constructed a cDNA encoding a 146-kDa GST-TFII-I fusion protein that is readily distinguished from the endogenous 120 kDa TFII-I. In addition to the GST moiety, p146 also contained the hexa-histidine tag, thus allowing rapid purification. Figures 1A and B show, respectively, the expression and purification of GST-TFII-I (p146), which was monitored both by silver staining (Fig. 1A) and by Western blot analysis with an anti-TFII-I antibody (Fig. 1B). A purified p146 preparation bound to the V Inr probe, resulting in a lower mobility shift than the native TFII-I (Fig. 1C). Most importantly, the Inr element binding specificity of the recombinant protein was similar to that of the native protein, as evidenced by the competition assay (Fig. 1C), and the DNA-protein complex was abrogated by anti-TFII-I antibody (data not shown). Finally, we tested the transcriptional properties of p146 in a TFII-I immunodepleted nuclear extract. p146 was fully capable of restoring V transcription in this extract and thus demonstrated functional properties similar to those of native TFII-I (Fig. 1D) (29). Furthermore, V transcription reconstituted with p146 was inhibited by preincubation of p146 with an oligonucleotide containing the wild-type V Inr sequence, which specifically binds p146 but not by preincubation with an oligonucleotide containing a mutant Inr sequence (data not shown), suggesting that the Inr binding capabilities of TFII-I were required for its transcription function in vitro. Thus, these results demonstrate that the eukaryotically expressed recombinant TFII-I behaves in a fashion similar to the native protein both in DNA binding and in in vitro transcription assays.

View larger version (32K): [in this window] [in a new window]   FIG. 1.   Eukaryotically expressed recombinant TFII-I has DNA binding and transcriptional properties similar to native TFII-I. Silver stain (A) and Western blot (B) analyses with the anti-TFII-I antibody comparing the relative mobility and purity of the native, purified TFII-I (lane 1) to those of the recombinant protein (p146) at all stages of purification (lanes 2 to 5). Mock lysate was derived from COS cell extracts after transfection with the vector alone. Positions of GST-TFII-I (p146) and native TFII-I (p120) are indicated. (C) EMSA with V Inr to demonstrate that the GST-TFII-I has a binding specificity similar to that of purified TFII-I. While GST cannot bind the V Inr element (lane 2), purified TFII-I and the recombinant TFII-I (GST-TFII-I) can bind to the V Inr element (lanes 3 and 4). The binding of p146 can be competed with a 15-fold excess of wild type (WT) but not with a mutant (Mut) V Inr containing oligonucleotide (lanes 5 and 6). (D) V transcription is observed in the mock-treated Jurkat nuclear extract (lane 1) but not in an anti-TFII-I antibody-treated Jurkat nuclear extract (-TFII-I; lane 2); it can, however, be reconstituted by adding back either GST-TFII-I (lane 3) or native TFII-I (lane 4).

Functional interactions of TFII-I with the V Inr element in COS cells. To ascertain the minimum promoter region that is required to mediate transcriptional response of ectopically expressed TFII-I (146 kDa) and to discriminate between its upstream-sequence-dependent and -independent (basal) functions, we initially generated two V promoter templates each fused to the luciferase gene. The first contained the full-length V promoter (476 to 9, VFL), and the second contained only 61 nucleotides from the start site (61 to 9, V61). Furthermore, sequences located 3' of the Inr were removed in both cases to minimize confusion due to TFII-I interactions with downstream promoter elements (3, 4). To show Inr-specific transcription functions of TFII-I in vivo at the V promoter, the Inr elements in the VFL and V61 minimal promoters were also mutated. The specific mutations introduced within the consensus Inr sequence were known to disrupt the basal Inr function (21). The schematic representation of these reporters is shown in Fig. 2A.

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Initiator-dependent transcriptional activation of the V promoter by TFII-I. (A) The architecture of full-length V (VFL-Luc) and truncated V luciferase (V61-Luc) promoter constructs with wild type (WT) or mutant (Mut) initiator sequences are shown on the left. The binding sites for various transcription factors are indicated. Deca denotes the conserved decamer sequence that is present in all V promoters (1). COS7 cells were transiently transfected with 1 µg of either the wild-type (lanes 3 and 4) or the mutant (lanes 5 and 6) VFL-Luc plasmids in the presence (lanes 4 and 6) or in the absence (lanes 3 and 5) of 350 ng of TFII-I expression plasmid pEBGII-I. For the V61-Luc wild-type (lanes 7 and 8) or mutant (lanes 9 and 10) plasmids, 75 ng was used for cotransfection in the presence (lanes 8 and 10) or absence (lanes 7 and 9) of 350 ng of pEBGII-I. Lanes 1 and 2 represent the cotransfection of control plasmids (PGL3-Basic) with or without pEBGII-I. To equalize the total amount of DNA, 350 ng of empty vector (pEBB) was added in lanes with TFII-I. The experiments represent five (for lanes 1 to 6) and three (for lanes 7 to 10) independent transfection assays. The fold activation was calculated by normalizing the basal promoter activity of either the full-length or the truncated promoter to 1. The absolute values were twofold lower for the V61-Luc promoter than the VFL-Luc promoter. (B) EMSA showing that V initiator mutation reduces the binding of TFII-I. Equal counts (approximately 60,000 cpm) of V wild-type (WT) (lanes 1 to 3) and V mutant (Mut) (lanes 4 and 5) probes were used per reaction. Lanes 2, 3, and 5 contained TFII-I. Lane 3 also contained affinity-purified anti-TFII-I antibody. (C) Western blot showing the expression of recombinant TFII-I (p146) in transient-transfection experiments. COS cell lysates (5 µg) from VFL-Luc only (lane 2) or VFL-Luc+pEBGII-I (lane 3) transfections were subjected to Western blot analysis with an anti-TFII-I antibody. Native TFII-I (p120) was used as a positive control (lane 1).

Isolation of the DNA binding fragment of TFII-I by limit proteolysis. Limited proteolysis is a classical biochemical method to identify structural domains within a large multidomain protein (11, 20, 23). To identify the DNA binding fragment of TFII-I, we employed limited proteolysis in order to separate the protease-resistant versus protease-sensitive domains and to select the potential DNA binding domain of TFII-I. Thrombin digestion of native, purified TFII-I was performed for various times, and the proteolyzed products were analyzed by DNA binding assays (Fig. 3). Thrombin treatment of native TFII-I prior to the DNA binding reaction resulted in a shifted complex with higher mobility compared to the untreated-gel-shifted complex in a time-dependent fashion (Fig. 3A). This complex was stable and protease resistant even after 1 h of thrombin treatment (data not shown). Moreover, when thrombin treatment was carried out subsequent to the DNA binding reaction, the DNA-bound TFII-I gave the same pattern as did the unbound TFII-I (Fig. 3B), suggesting that DNA binding does not induce any significant conformational changes.

View larger version (56K): [in this window] [in a new window]   FIG. 3.   Protease-resistant DNA binding fragment of TFII-I. (A) EMSA demonstrating that the thrombin-resistant fragment of TFII-I exhibits DNA binding activity. Purified TFII-I (lane 1) was treated with 1 CLU of thrombin at room temperature for 0 to 40 min (lanes 2 to 5). At the end of the digestion the reaction mixture was incubated with an oligonucleotide containing the wild-type V Inr at 30°C for 20 min. (B) DNA binding by TFII-I does not induce conformation changes. Purified TFII-I was incubated with the V Inr containing oligonucleotide at 30°C for 20 min and digested with 1 CLU of thrombin at room temperature either for 0 (lane 2) or 30 (lane 3) min.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   Protease treatment cleaves TFII-I into two fragments. (A) Western blot analysis with anti-TFII-I antibody showing the formation of a 70-kDa fragment (p70) by treating either GST-TFII-I (lanes 1 to 4) or native TFII-I (lanes 5 to 8) with 1 CLU of thrombin for 30 min at room temperature. The formation of 100- and 70-kDa products are indicated. (B) Western blot analysis with anti-TFII-I antibody to show the formation of a protease-resistant 102-kDa polypeptide (arrow) by treating either native TFII-I (lanes 1 to 3) or GST-TFII-I (lanes 4 to 6) with V8 protease at room temperature for 0 to 30 min. (C) Silver staining of native homogeneous TFII-I before (lane 1) and after (lane 2) thrombin digestion for 20 min. The 100-, 70-, and 43-kDa fragments are indicated. (D) Diagram of TFII-I. The positions of the direct repeats (R1 to R6) are indicated. The arrow indicates the major thrombin cleavage site at aa 677 between repeats 4 and 5 (based on N-terminal sequencing).

Generation of recombinant p70 in bacteria. To directly demonstrate that p70 indeed contains the DNA binding domain of TFII-I, a cDNA encoding aa 1 to 735 was subcloned into a bacterial expression vector that adds a hexa-histidine tag to the N terminus of the protein. Although the p70 generated by proteolysis contained 677 aa, for the sake of cloning convenience our recombinant p70 construct contained 58 extra aa at the C-terminal end. The expressed protein was purified from crude bacterial lysate by binding to an Ni²⁺-agarose column. The purified recombinant p70 protein (Fig. 5A, lane 1) was subjected to Western blot analysis and compared to a native 120-kDa TFII-I (lane 2). To demonstrate that recombinant p70 contained the DNA binding domain, it was compared by EMSA to the native TFII-I before and after thrombin cleavage (Fig. 5B). The mobility of the p70 fragment (lane 3) generated from thrombin cleavage of the native TFII-I (lane 2) was similar to the mobility of the recombinant p70 (lane 4). Furthermore, the p70 gel-shifted complex was completely abrogated in the presence of an affinity-purified anti-TFII-I antibody (Fig. 5C, compare lanes 3 and 4) and is specific since it could be competed strongly by an oligonucleotide containing the wild-type V Inr sequence (lane 5) but very weakly by an oligonucleotide that contains the mutated V Inr element (lane 6). We concluded that p70 contains the DNA binding domain of TFII-I and is sufficient to impart Inr specificity and that the C-terminal portion of TFII-I is dispensable for its DNA binding activity.

View larger version (44K): [in this window] [in a new window]   FIG. 5.   Expression and DNA binding specificity of recombinant p70. (A) Western blot analysis with anti-TFII-I antibody of a native purified TFII-I (lane 1) and the bacterially expressed recombinant p70 (lane 2). (B) EMSA demonstrates that the native TFII-I (lane 2), when treated with thrombin, generated a complex (lane 3) that has mobility comparable to the mobility of the complex derived from the recombinant p70 (lane 4). (C) EMSA demonstrates that purified TFII-I (lane 2) and recombinant p70 (lane 3) can bind to the V Inr element comparably. Anti-TFII-I antibody blocked the binding of recombinant p70 (lane 4), and the binding of recombinant p70 can be competed with a 15-fold molar excess of wild type (lane 5), but not with a mutant Inr-containing oligonucleotide (lane 6).

p70 lacks independent activation potential and inhibits wild-type TFII-I function. To determine the transcription functions of recombinant p70, we employed the transient-transfection assay in COS7 cells. For ectopic expression of p70, we used a construct that would produce a GST-p70 (96 kDa) fusion protein (Fig. 6A). Whereas the wild-type TFII-I stimulated the VFL promoter by about fourfold (Fig. 6B, compare lanes 3 and 4), p70 alone did not activate this promoter (compare lanes 3 and 5). Cotransfection of both TFII-I and p70 resulted in a complete abrogation of the V transcriptional activation (compare lanes 4 and 6). These results suggested that p70 lacks the transcription activation potential and that the C-terminal 280 aa might constitute or contain an activation domain. In order to test this hypothesis, we fused a heterologous activation domain of the yeast activator protein GAL4 (activation domain II, aa 768 to 881 [29, 45]) to p70 (p70-GAL4ADII [Fig. 6A]). As seen in Fig. 6C, while p70 failed to stimulate the V promoter on its own (compare lanes 3 and 5), the p70-GAL4ADII fusion protein stimulated the V promoter nearly as much as had the wild-type TFII-I (compare lanes 4 and 6), suggesting that p70 indeed lacked an activation domain. As expected, unlike p70 p70-GAL4ADII did not inhibit the wild-type TFII-I function when coexpressed (data not shown). Furthermore, the transcriptional activation by p70-GAL4ADII required an intact Inr element (Fig. 6D) since significant activation was only observed with a wild-type V promoter (compare lanes 3 and 4) but not with an Inr mutant V promoter (compare lanes 5 and 6). Finally, to demonstrate that the various TFII-I expression constructs do express the respective proteins in comparable amounts in vivo under our transfection conditions, a Western blot was performed with the lysates expressing the empty vector (lane 1), the V promoter alone (lane 2), or the V promoter with GST-TFII-I (lane 3), GST-p70 (lane 4), GST-TFII-I+GST-p70 (lane 5), or p70-GAL4ADII (lane 6) and probed with an anti-TFII-I antibody (Fig. 6E). The various TFII-I derivatives are expressed at comparable levels under our assay conditions either individually (lanes 3, 4, and 6) or in combination (lane 5). Taken together, these results suggest that (i) TFII-I is composed of essentially two domains, the N-terminal portion containing a compact, protease-resistant DNA binding domain (p70) and the C-terminal portion required for transcriptional activation functions; (ii) p70, independently, does not activate the V promoter in transient-transfection assays and behaves as a dominant negative when coexpressed with the wild-type TFII-I; and (iii) when fused with a heterologous activation domain of yeast activator GAL4, p70 shows an Inr-specific transcription function comparable to that of the wild-type TFII-I. Therefore, the Inr-dependent function of TFII-I is solely dictated by its DNA binding specificity.

View larger version (26K): [in this window] [in a new window]   FIG. 6.   p70 lacks independent transactivation function and behaves as a dominant negative in transient-transfection assays. (A) Diagrams showing p70 and p70 fused to the GAL4 activation domain II (p70-GALADII). (B) COS7 cells were transiently transfected with either 1 µg of PGL3-Basic as a control (lanes 1 and 2) or 1 µg of VFL-Luc promoter (lanes 3 to 6) in the presence of 350 ng of pEBGII-I (lanes 2 and 4), 500 ng of p70 (lane 5), or both (lane 6). (C) The transactivation potentials of p70 can be rescued by fusion of a heterologous GAL4 activation domain to p70. Lanes 1 to 5 are the same as in panel B. Lane 6 contains 250 ng of the p70-GAL4ADII expression plasmid. The activation by p70-GAL4ADII is comparable to that of wild-type TFII-I (lane 4). (D) p70-GAL4ADII activates V promoter in an initiator-specific manner. COS7 cells were transiently transfected with 1 µg of either wild-type (lanes 3 and 4) or mutant (lanes 5 and 6) VFL-Luc promoter in the presence (lanes 4 and 6) or absence (lanes 3 and 5) of 250 ng of p70-GAL4ADII. Lanes 1 and 2 represent the cotransfection of PGL3-Basic with or without p70-GAL4ADII. The total amounts of DNA in experiments as indicated in panels B, C, and D were equalized by the addition of empty vector (pEBB), and the results represent an average of three independent experiments. The fold activation was calculated by normalizing the VFL-Luc promoter activity to 1. (E) Western blot analysis of lysates from transfection assays to show the comparable expression of various TFII-I derivatives in our transient-transfection assays. Equal amounts of lysate (5 µg of total protein) transfected with PGL3-Basic (lane 1), VFL-Luc alone (lane 2), or VFL-Luc plus GST-TFII-I (lane 3), GST-p70 (lane 4), GST-TFII-I+GST-p70 (lane 5), or p70-GAL4ADII (lane 6) were Western blotted with anti-TFII-I antibody.

View larger version (10K): [in this window] [in a new window]   FIG. 7.   TFII-I derivatives do not function through a TATA-containing but Inr-lacking promoter. COS7 cells were transiently transfected with 200 ng of pFR-Luc alone (lane 1) or cotransfected with 350 ng of plasmids expressing GAL4DBD-GAL4ADII (GAL4; lane 2), TFII-I (lane 3), p70 (lane 4), p70-GAL4ADII (lane 5), or GAL4DBD (aa 1 to 147) fused to the C-terminal 280 amino acids of TFII-I (GAL4DBD-C280; lane 6). The total amounts of transfected DNA in all lanes were equalized by the addition of empty vector (pEBB). The results represent three independent transfection assays. The fold activation was relative to the pFR-Luc activity that was taken as 1.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Since the discovery of the initiator as an autonomous core promoter element (41), the mechanism of action of this element has remained unclear and controversial. This reflects in part both the low efficiency of Inr function, relative to strong TATA elements, and the lack of a completely defined in vitro system for Inr function. It is clear, however, that basal Inr-directed transcription requires some factors not required for basal TATA-directed transcription in vitro and that a subset of these factors is required for Inr function in the context of a TATA- and Inr-containing promoter (36). The demonstration of several Inr or Inr-region interacting factors in vitro has further complicated the issue and leaves open the possibility of diverse Inr-mediated transcription initiation pathways (43, 49). To resolve these problems, systematic functional assays, in addition to the in vitro assays, must be undertaken to address the physiological significance of Inr-dependent interactions. Toward this end, we have undertaken an in vivo analysis of TFII-I, an Inr-binding protein, and show here an essential role for ectopic TFII-I in transcription of the naturally TATA Inr⁺ V 5.2 promoter (1, 2).

The in vivo functions of ectopic TFII-I were first shown by us with the potent TATA- and Inr-containing AdML model promoter (39). However, for natural promoters the biochemical interactions and cofactor requirement for Inr function in the presence of a TATA box might be different than in its absence, further suggesting the importance of the promoter context on Inr function (36). Hence, clear demonstration of TFII-I function via an Inr element in a natural TATA Inr⁺ promoter was necessary. The V promoter not only allowed us to directly address the role of TFII-I in the context of a TATA-less promoter, but it also allowed us to preserve the natural promoter context which could be critical for appropriate expression of several TATA Inr⁺ genes (9, 12, 31). As shown by complementary functional assays with nuclear extracts, which were expected to contain all of the factors necessary for Inr function, TFII-I is required for transcription of the V promoter. This was demonstrated by the ability of an antibody directed against the cDNA-encoded protein to selectively inhibit transcription of the V promoter and by the ability of a recombinant TFII-I, expressed in and purified from mammalian cells, to restore the V transcription in immunodepleted nuclear extracts.

Together with the in vitro data, here we show that ectopically expressed TFII-I markedly stimulates the expression of the V promoter in in vivo (transient-transfection) assays. Given the fact that promoter truncation only affects the expression of the V promoter by twofold, the expression of the promoter, even in the presence of TFII-I, reflects largely basal-level (activator-independent) transcription. The V promoter contains an upstream E-box motif (consensus: CAGGTG). However, in functional assays, this E-box element does not appear to contribute significantly toward TFII-I-dependent transcriptional activation. Because this consensus differs from that of the adenovirus major-late-promoter E box (consensus: CACGTG) through which TFII-I functions (39), it is possible that TFII-I activates transcription only through specific E boxes. Although the V promoter also contains sequences for inducible and tissue-type restricted activators (e.g., NFB and Pu.1), these factors are largely absent from COS cells. Thus, it is not too surprising that deletion of upstream sequences in COS cells does not result in a significant reduction in promoter activity. However, V belongs to a complex family of core promoters (16) that could potentially have activator elements immediately adjacent to the Inr element. In the latter scenario, even the 61 "core" promoter might contain unidentified activator sequences. Added to this complexity is the fact that the absolute levels of expression of the V promoter constructs in transient assays are rather low; thus, our attempts to analyze the transcripts by RNA assays have been largely unsuccessful (data not shown). However, it is clear that the V promoter requires TFII-I for efficient expression both in vitro and in vivo, regardless of whether the readout reflects activator-dependent or -independent transcription.

It is also clear from our analysis that an intact Inr element is required for TFII-I functions, since either mutations in the Inr (V) or the lack of an Inr element (but the presence of a TATA box, pFR-Luc) resulted in a lack of transcriptional activation by any derivatives of TFII-I. That TFII-I binding to the V Inr element is tightly correlated with transcriptional functions in vivo is also consistent with the notion that TFII-I binds and functions directly through the V Inr element. Although both our in vitro and especially our in vivo data are most consistent with direct involvement of TFII-I via the Inr element, in the absence of altered specificity mutants of TFII-I we cannot at present unambiguously demonstrate that TFII-I directly functions through the Inr element in vivo. Furthermore, it remains a formal possibility that the lack of TFII-I responsiveness of the Inr mutant promoters might be due to the fact that these mutations render the promoters completely inactive, especially in the absence of an RNA assay that could detect them. However, we do not favor this notion because the transcriptional levels from the mutant promoters, although 40% lower than those of the wild-type promoters, are still readily measurable.

Creation and use of a dominant negative mutant of an activator protein are a powerful approach to assigning a direct and unambiguous functional role in vivo. For a DNA binding transcription factor this could occur if the DNA binding domain is left intact but the activation domain is deleted or mutated (18, 19). Thus, as a first step to achieving this, we biochemically identified and isolated the DNA binding domain of TFII-I, reasoning that if this fragment or domain retained specific DNA binding ability but lacked the activation function then it might behave as a dominant negative mutant of wild-type TFII-I in vivo. The p70 mutant of TFII-I demonstrated Inr binding specificity in vitro but lacked any detectable transcription functions in transient reporter assays. As expected given these properties, p70 behaved as a dominant negative mutant of wild-type TFII-I when both proteins are coexpressed. Furthermore, ectopically expressed p70 also inhibited the basal V promoter, suggesting that it competes for the low levels of endogenous TFII-I. Despite the fact that this inhibition is only 30 to 40%, it is comparable to that obtained with Inr mutation. However, the exact mechanisms of the dominant negative function of p70 are not yet clear. Although the binding of p70 to the Inr is only marginally better than the wild-type TFII-I, p70 in functional assays totally inhibits the wild-type TFII-I function. Thus, a simple model of competition for promoter occupancy might not explain such dramatic effects and might involve titration of cofactors or heterodimerization between the wild-type and the mutant proteins, especially in vivo. Regardless of the exact mechanism of inhibition, it is clear that p70 lacks an activation domain because fusion of the GAL4 activation domain to p70 rescued its transcriptional potentials. Thus, by inference, the C-terminal 280 aa of TFII-I must contain or be part of an activation domain. Surprisingly, this activation domain is not required for Inr-specific transcription since the GAL4 activation domain imparts Inr function and suggests that the Inr specificity is largely dictated by the DNA binding domain of TFII-I. On the other hand, when the C-terminal 280 aa was fused to the DNA binding domain of GAL4 (aa 1 to 147), it failed to impart any detectable transcriptional responses from a promoter that contained five GAL4 binding sites upstream of a TATA box. In the same assay, the GAL4 activation domain, fused to its own DNA binding domain, stimulated the promoter significantly. Because this promoter contained a TATA box but lacked an Inr element, it is possible that the activation function of the C-terminal 280 aa may only be detectable in the presence of an Inr as has been shown for other activation domains (7, 27). However, it is also possible that this C-terminal domain of TFII-I is necessary but not sufficient for activation function and that it requires other portions of TFII-I for appropriate transcriptional responses.

What is the mechanism by which TFII-I stimulates Inr-dependent transcription either in the presence of a TATA box (39) or in its absence (this study)? Because TFII-I appears to have two distinct and separable domains, could it act as a more conventional activator protein even when bound to the Inr element? Moreover, the activation domain of GAL4 can function in an Inr-specific fashion when fused to the DNA binding domain of TFII-I. The GAL4 activation domain is known to target components within the basal machinery, most notably the TATA-binding protein and TFIIB, and it might help recruit the transcriptional machinery (reviewed in reference 35). It is likely that the C-terminal domain of TFII-I interacts with the same components of the basal machinery, although this domain may not, independently, direct significant transcription. However, we do not favor the notion that TFII-I behaves as a conventional "activator" protein. Rather, we favor a model in which TFII-I possesses one C-terminal activation domain and two separable DNA binding domains embedded within the N-terminal domain: one specific for activator function and the other specific for basal function. Either of these DNA binding domains might function independently with the C-terminal activation domain and, depending on the usage of these DNA binding domains, TFII-I can behave either as an activator or as a basal factor. Further structure-function analysis of TFII-I and identification of its targets within the basal machinery will reveal the critical determinants and will reveal the mechanisms of action of this intriguing transcription factor.