INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Eukaryotic gene expression is controlled by a combination of effects from various cis-elements present in the promoter and enhancer regions. One such cis-element is the E box (CANNTG), which is recognized by a variety of basic helix-loop-helix (bHLH)¹ transcription factors. Two types of E box, depending upon the nature of the two central nucleotides, are described (1). Whereas the E box with the sequence CAGCTG is referred to as class A, the E box with the sequence CACGTG belongs to class B. The functions of the E box are very divergent and to a large extent dependent upon the transcription factors that bind to it. These include neurogenesis, myogenesis, sex determination, T-cell/B-cell, and pancreatic specific gene expression, as well as cell proliferation and differentiation (reviewed in Ref. 2). The selective binding mechanism by which multiple factors bind to the same target site (class A or B) is not clear. It may involve competition among the factors influenced by sequences flanking the consensus CANNTG or interactions with other proteins binding to adjacent sites (3).

Upstream stimulatory factor (USF) is a ubiquitous transcription factor (4) belonging to the class B proteins that also include Myc (5), Max (6), Mad (7), Mxil (8), TFEB (9), TFE3 (10), and TFEC (11). USF was first identified for its stimulation of transcription from the adenovirus late promoter (13) and was purified from HeLa cells as two polypeptides, USF1 (43 kDa) and USF2 (44 kDa) (14). USF1 and USF2 bind to the E box (CACGTG) as homo- and heterodimers, and their ratios vary in different cell types (4). Although the exact biological roles of USF1 and USF2 are not fully understood, it has been reported that they play a critical role in both the basal (15-19) and signal-induced (4, 20-22) expression of cellular genes.

Previously (23) we have shown that the promoter activity of human transcobalamin II (TC II), a plasma transporter of cobalamin (24), is relatively weak and controlled positively by a distal GC box and negatively by a proximal GC/GT overlapping box. In the present study, we have identified a 69-bp sequence (581/513) from the distal region of the TC II promoter (25) that did not contain an obvious TATA box or a known Inr element but possessed a high promoter activity in transfected cells in an orientation-independent manner. The bidirectional promoter activity was due to interactions between Sp1/Sp3 and USF1/USF2 that bound to GC box and E box, respectively. Furthermore, a 29-bp sequence (541/513) containing only the E box was sufficient by itself to mediate transcription in the absence of other cis-elements. This finding suggests that TATA/Inr promoters may use an E box as a core element to direct basal transcription.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of Promoter-CAT Reporter Plasmids-- Promoterless plasmid, pCAT-Basic (pCAT-B) (Promega, Madison, WI), was used for the preparation of the CAT fusion reporter constructs. Various truncated TC II-promoter fragments were generated by polymerase chain reaction (PCR) and inserted into the pCAT-B vector at a PstI site upstream from the CAT gene. A total of seven promoter fragments were amplified, and the sequences of each pair of primers used for PCR are shown in Table I. The DNA sequence of each of the promoter fragments was confirmed by sequencing prior to transfection. For promoter fragments CI (25-bp) and CII (29-bp), double-stranded (DS) oligonucleotides corresponding to the respective sequences (Table II) were synthesized and ligated at a PstI site in front of the CAT gene in the pCAT-B vector. In addition, fragments F2 and CII were also cloned into the pCAT-Promoter (pCAT-P)(Promega) vector at a site about 2.8 kilobase pairs upstream of the start site of the CAT gene.

Site-directed Mutagenesis-- The substitution mutagenesis of the GC or E box in the promoter fragment was generated using the QuickChange Site-directed mutagenesis method (Stratagene, La Jolla, CA). For each mutagenesis a pair of overlapping oligonucleotides with desired mutations was commercially synthesized, and the sequences are listed in Table II. The mutations generated were confirmed by DNA sequencing.

Cell Culture and Transient Transfection-- The human colon epithelial cell line Caco-2 cells were obtained from American Type Culture Collection (Rockville, MD) and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells were plated at a density of 1.5 × 10⁶ cells/100-mm dish the day before transfection and transfected by the calcium phosphate precipitation method (26). The transfection was performed using 15 µg of promoter-CAT fusion plasmid and 5 µg of an internal control plasmid pSV--galactosidase (Promega) for monitoring the transfection efficiency. In addition, promoterless vector pCAT-B and promoter-containing plasmid pCAT-P were also transfected in each experiment as a negative and positive control, respectively. Each transfection was carried out at least three times.

-Galactosidase and CAT Assays-- After 46 h of transfection, cells were harvested, and extracts were prepared for -galactosidase and CAT assays. -Galactosidase activity was measured according to Herbomel et al. (27), and CAT activity was determined by the method of Gorman et al. (28) using [¹⁴C]chloramphenicol as a substrate. The acetylated products were separated from the unacetylated products by either thin layer chromatography or liquid scintillation counting as described by Seed and Sheen (29). CAT activities were normalized to -galactosidase activities for the variations of transfection efficiency.

Preparation of Nuclear Extracts-- Nuclear extracts were prepared from Caco-2, HeLa, and K-562 cells essentially as described by Dignam et al. (30). All buffers contained protease inhibitors including phenylmethanesulfonyl fluoride (1 mM), leupeptin (2 µg/ml), antipain (10 µM), and benzamidine (1 mM). Protein concentration of the nuclear extract was determined by the Bio-Rad protein assay method using bovine serum albumin as a standard (31).

Electrophoretic Mobility Shift Assay (EMSA)-- The promoter fragment C (581/513) was labeled with [³²P]dCTP using the Klenow fragment of DNA polymerase. DS oligonucleotide CI or CII (Table II) was labeled at the 5' termini with [-³²P]ATP using T4 polynucleotide kinase. Labeled probe (~2 × 10⁴ cpm) was incubated for 15 min at 22 °C with 2.2 µg of nuclear extract in 10 µl of reaction buffer (10 mM Tris-HCl, pH 7.5, 4% glycerol, 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, and 50 µg/ml poly(dI·dC)·poly(dI·dC)). For competition experiments the nuclear extract was incubated with the indicated concentrations of DS oligonucleotide competitor (Table II) at 22 °C for 5 min prior to incubation with the probe. For immunosupershift assay, the nuclear extract was preincubated with 1.5 µg of affinity purified rabbit polyclonal antibody against USF1, USF2, or c-Myc (Santa Cruz Biotechnology, Inc, Santa Cruz, CA) at 22 °C for 30 min prior to incubation with the probe. The reaction mixture was then subjected to 4% polyacrylamide gel electrophoresis in 0.5× TBE buffer (44.5 mM Tris-HCl, 44.5 mM boric acid, and 1 mM EDTA) at 100 V. The protein-DNA complexes were visualized by autoradiography.

DNase I Footprinting Analysis-- The coding strand of the promoter fragment C was 3'-end-labeled by [³²P]dCTP using the Klenow fragment of DNA polymerase and gel-purified. The end-labeled probe (~1 × 10⁴ dpm) was incubated for 10 min on ice with 20 µg of nuclear extract in 50 µl of buffer (25 mM Tris-HCl, pH 8.0, 50 mM KCl, 6.25 mM MgCl₂, 0.5 mM EDTA, and 10% glycerol). After the incubation, 50 µl of 5 mM CaCl₂, 10 mM MgCl₂ solution was added to the reaction followed by digestion with 0.4-2 units of DNase I (Promega) at 22 °C for 2 min. The digestion was then terminated by adding 90 µl of stop solution (200 mM NaCl, 30 mM EDTA, 1% SDS, and 100 µg/ml yeast tRNA) and subjected to phenol/chloroform extraction. After precipitation with ethanol, the DNA pellet was resuspended in 4 µl of the loading buffer, heated at 95 °C for 2 min, and subjected to electrophoresis in a 6% polyacrylamide sequencing gel.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

A 69-bp Sequence Derived from the 5'-Region of the TC II Gene Activates Transcription in an Orientation-independent but Position-dependent MannerOur previous 5'-deletion studies (23) have demonstrated that the transcriptional activity of the TC II-promoter fragments, including the longest (1014 to +34), is very weak in Caco-2 cells. To identify regions that may potentially have higher promoter activity, four promoter fragments (F1-F4) generated by PCR were cloned into a promoterless CAT reporter vector, pCAT-B. Upon transient transfection in Caco-2 cells (Fig. 1A), only the fragment F2-(746/513) possessed promoter activity above the background activity produced by the promoterless pCAT-B vector. The promoter activity of this fragment (20-fold) was nearly 2.5-fold higher than that obtained using the pCAT-P vector (8.5-fold) that contained the SV40 promoter upstream from the CAT gene. In addition, the promoter activity of the F2 fragment was orientation-independent (Fig. 1B) as the fragment F2 in a reverse orientation (513/746) revealed a similar level (19-fold) of transcriptional activity.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Deletion analysis of the 5'-flanking region of the human TC II gene. A series of truncated promoter fragments were constructed in front of the CAT gene in a promoter-less vector pCAT-B or at a remote position relative to the transcriptional start site of the CAT gene in a promoter-containing vector pCAT-P. The fusion constructs and the vector pCAT-B or pCAT-P were transiently transfected in Caco-2 cells. The relative CAT activity was expressed as fold activation over the activity of the vector pCAT-B (closed bars) or pCAT-P (open bars). The data shown are means ± S.D. from five transfections. A, relative CAT activities of fragments F1 to F4. The arrows in the linear diagram of the promoter region indicate the windows of the transcription start sites. The nucleotide numbers are relative to the G(1) of the putative translation initiation codon, ATG. B, CAT activities of fragments A, B, and C derived from fragment F2 showing that they function in an orientation-independent but position-dependent manner.

In order to identify the elements within the 234-bp fragment that were responsible for its high bidirectional transcriptional activity, this fragment was further dissected into three regions (A, B, and C) (Fig. 1B). The 5'-terminal region A (58 bp, 746/689), which contained an inverted CCAAT element and a potential AP2-binding site, led to a 3-fold activation of the CAT activity in both orientations, accounting for about 16% of the activity of the fragment F2. The region B (119 bp, 688/570) which contained several potential cis-elements, such as two AP2 sites, two Myb sites, and four inverted GA or GT boxes (GGGA/TGGG), did not reveal any promoter activity in either orientation. However, the 3'-terminal region C (69 bp, 581/513) had a high transcriptional activity (~15-fold) in both orientations, corresponding to about 75% of promoter activity of the full-length F2 fragment.

Although the fragment F2-(746/513) or fragment C-(581/513) possessed relatively high promoter activities, the activity was masked when a downstream region (513/+183) was included (23). The strong reduction of the transcriptional activity could be due to either a distance effect or to the presence of a GC/GT box acting as a negative element in the downstream region (23). To distinguish between these two possibilities, the fragment F2 or C was constructed into a pCAT-P vector containing the SV40 promoter at a remote position, nearly 2.8 kilobase pairs upstream of the transcriptional start site of the CAT gene. As shown in Fig. 1B, very little or no activation was observed using either of the two constructs, F2-(581/513)-P or C-(581/513)-P. These results suggested that the fragment F2 or C functioned in a distance-dependent manner. Taken together, these results suggested that the 69-bp sequence was mainly responsible for the transcriptional activity of the 234-bp fragment and that it functioned in an orientation-independent but a distance-dependent manner.

The 69-bp Sequence Contains Two Functional cis-Elements, a GC box and an E box (CACGTG)-- Inspection of the sequence of the 69-bp region revealed a potential Myb site, a GC box with one mismatch, and two types of E box, CAGCTG (class A) and CACGTG (class B). To identify functional cis-elements in the 69-bp region, DNase I footprinting analysis was performed using nuclear extracts from Caco-2 and HeLa cells. Two protected regions, CI and CII, were revealed using both cellular nuclear extracts (Fig. 2A). Region CI-(578/559) included a Myb-binding site and a previously described (23) GC box, and region CII-(532/519) contained a 14-bp palindromic sequence with an E box (CACGTG) of class B in the center of the palindrome (Fig. 2B).

View larger version (38K): [in this window] [in a new window]   Fig. 2.   Mapping the cis-elements of the 69-bp region by DNase I footprinting. A, DNase I footprinting showing the two protected regions at CI-(559/578) and CII-(519/532). 32P-Labeled promoter fragment C was digested with 1 (lanes 1 and 5), 2 (lanes 2 and 6), 0.4 (lanes 3 and 7), and 0.6 (lanes 4 and 8) units of DNase I in the presence or absence of nuclear extracts as indicated. DNA sequencing ladder (lane 9) was used as a molecular weight marker. B, nucleotide sequence of the fragment C showing the two protected regions and the potential cis-elements.

View larger version (47K): [in this window] [in a new window]   Fig. 3.   Identification of the nuclear proteins binding to the protected regions, CI and CII, by EMSA. A, 32P-labeled DS oligonucleotide CI (Table II) containing a GC box with one nucleotide mismatch and a potential Myb-binding site was incubated with (lanes 2-6) or without (lane 1) nuclear extracts from Caco-2 cells. The nuclear extracts were preincubated with binding buffer alone (lane 2) or with unlabeled DS oligonucleotides as indicated (lanes 3-6). B, 32P-labeled DS oligonucleotide CII (Table II) containing a 14-bp palindrome sequence with an E box in its center was incubated with (lanes 2-5) or without (lane 1) Caco-2 cell nuclear extracts. The nuclear extracts were preincubated with buffer alone (lane 2) or with unlabeled DS oligonucleotide competitors (lanes 2-4) as indicated.

Both the GC Box and the E Box Are Required for the Transcriptional Activity of the 69-bp Promoter Region-- To evaluate the role of the GC box and the E box on the promoter activity of the 69-bp region, the same mutations that eliminated nuclear protein binding (Fig. 3) were introduced into the plasmid construct C-(581/513). Fig. 4 shows the CAT activity of the wild-type or the mutant constructs transfected in Caco-2 cells. Mutations in the GC box resulted in about 50% loss of the CAT activity of the 69-bp fragment, whereas mutations in the E box alone or in both the GC box and the E box diminished the transcriptional activity by about 90%. These results indicated that both the GC box and the E box were required for the full transcriptional activation of the 69-bp fragment. However, the relative contributions of these two cis-elements were different. The presence of the E box was essential for the transcriptional activity contributed by the GC box but not the other way around. On the other hand, the presence of the GC box enhanced the transcriptional activity of the E box by 2-fold, suggesting that there could be functional physical interactions between the nuclear proteins that bind to the two sites. In order to test this possibility, gel-shift analysis was carried our using the 69-bp fragment containing both the GC box and the E box.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Effects of mutations in the GC box and/or the E box on the promoter activity of the 69-bp fragment. Four promoter-CAT fusion constructs with wild-type sequence or mutations in the GC box, E box, or both the GC box and the E box were transiently transfected into Caco-2 cells. The relative CAT activities corresponding to each of the constructs were obtained from three independent transfections, and the values are given as means ± S.D. The CAT activity of the wild-type constructs minus the background activity obtained from the vector alone is taken as 100% and used as a reference to normalize the CAT activity of each mutant construct. The sequence of the wild-type and the mutant GC box and E box (indicated by asterisks) is shown.

Nuclear Proteins Bound at the GC Box and the E Box Interact Physically-- When the labeled 69-bp promoter fragment was allowed to bind to the nuclear extracts from Caco-2 cells, at least five DNA-protein complexes were revealed (Fig. 5, lane 1). When competed with excess unlabeled oligonucleotide CI (lane 3) or Sp1 (lane 6) which contained the GC box, the complexes I, II, III, and IV were eliminated. When the competition was carried out using unlabeled oligonucleotide CII carrying the E box, the complexes I, II, and V disappeared (lane 4). Competition with oligonucleotide CI-M containing mutations in the GC box (lane 2) or CII-M containing mutations in the E box (lane 5) did not affect the formation of the complexes I-V. These results suggested that the formation of the complexes I and II was the result of physical interactions between nuclear proteins bound to both the GC box and the E box since both the complexes were unable to form in the presence of either unlabeled oligonucleotide CI (lane 3) or CII (lane 4).

View larger version (58K): [in this window] [in a new window]   Fig. 5.   EMSA showing physical interactions between nuclear proteins bound to the GC box and the E box in the 69-bp fragment. The 32P-labeled DNA fragment C-(581/513) was incubated with Caco-2 cell nuclear extracts in the absence (lane 1) or presence (lanes 2-6) of competing DS oligonucleotides as indicated.

The Oligonucleotide CII (29-bp) Is Able to Direct Bidirectional Transcription in Three Types of Cells-- Initially we tested by EMSA whether the 29-bp fragment is able to bind to nuclear factors in K-562 and HeLa cells. As shown in Fig. 6, two DNA-protein complexes were formed when the ³²P-CII fragment was incubated with nuclear extracts from either HeLa (Fig. 6A, lane 1) or K-562 (Fig. 6B, lane 1) cells. This pattern was similar to that obtained using nuclear extracts from Caco-2 cells (Fig. 3B). The formation of the two complexes were eliminated when competed with excess unlabeled oligonucleotide CII (Fig. 6, A and B, lanes 2-4). In contrast, the oligonucleotide CII-M was unable to compete for the formation of the two complexes (Fig. 6, A and B, lanes 5-7). These results suggested that nature of the nuclear proteins binding to the 29-bp was similar or identical in all the three cell lines tested and that the binding was specific to the E box.

View larger version (49K): [in this window] [in a new window]   Fig. 6.   EMSA of the 29-bp sequence using nuclear extracts from HeLa and K-562 cells. The 32P-labeled DS oligonucleotide CII-(541/513) was incubated with HeLa (A) and K-562 (B) nuclear extracts and subjected to electrophoresis without (lane 1) or with (lanes 2-7) increasing amounts of unlabeled DS oligonucleotides as indicated.

USF Binds to the E box-- At least nine members of the bHLH/LZ family of transcription factors bind either as hetero- or homodimers to the E box core sequence (CACGTG). Whereas the c-Myc, Max, Mad, and Mxil are involved in the Myc network in controlling cell proliferation and differentiation (32), the others, TFEB, TFE3, and TFEC belong to the Mit subfamily (33) and are involved in the regulation of immunoglobulin heavy chain and insulin gene expression, and USF which is ubiquitously expressed (4) stimulates transcription of many genes (15-22). Thus, we chose USF and c-Myc as the first candidates to examine their binding to the E box.

View larger version (50K): [in this window] [in a new window]   Fig. 7.   Immunosupershift analysis of the 29-bp sequence showing the binding of USFs to the E box. The 32P-labeled DS oligonucleotide CII-(541/513) was incubated with Caco-2 (A) or HeLa (B) cell nuclear extracts. The nuclear extracts were preincubated with binding buffer alone (A and B, lane 1) or with the indicated antibodies (A and B, lanes 2-5).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The present study has provided some insights into the basal and activator-dependent transcription using a human TC II promoter fragment that lacked both a TATA box and an Inr element. By deletion and transient transfection studies, we have demonstrated that a 69-bp DNA sequence from the human TC II promoter possesses bidirectional promoter activity but not an enhancer activity (Fig. 1B, fragment C). The 69-bp fragment did not contain an obvious TATA box or a known Inr element but contained a GC box with one mismatch and a 14-bp palindromic sequence (TGCTCACGTGACCA) with an E box (underlined) in its center. These observations raised an interesting issue as to how the 69-bp DNA fragment without a TATA box and Inr element functioned as a promoter.

Our EMSA experiment (Fig. 3) has shown that both the GC box and the E box were functional in binding to nuclear factors, and immunosupershift analysis further demonstrated that the GC box interacted with Sp1 and Sp3 (Fig. 3, Ref. 23), whereas the E box was recognized by both USF1 and USF2 (Fig. 7). Site-directed mutagenesis (Fig. 4) demonstrated that both the GC and the E box were required for the full promoter activity of the 69-bp fragment. However, their individual contribution toward the promoter activity was not equivalent. Mutations in the GC box reduced the promoter activity by 50%, whereas mutations in the E box alone or in both the E box and the GC box resulted in about 90% reduction of the promoter activity. These results implied that the E box was required for both the GC-dependent and -independent promoter activity of the 69-bp fragment, and there was potential cooperative interactions between nuclear factors that bound to these two cis-elements. Direct evidence for physical interactions between Sp1 or Sp3 and USFs was provided by EMSA (Fig. 5). The essential role of the E box in mediating transcription was further demonstrated by transfection of the fusion plasmid containing the DS oligonucleotide II (29-bp) which contained only the E box in three different cell lines (Table III).

The observation that a short (29-bp) DNA fragment containing only one recognizable cis-element, an E box, is sufficient to direct bidirectional transcription efficiently is very interesting. In general, a eukaryotic core promoter contains either a TATA box or an Inr element that are recognized by the TATA-binding protein (TBP), a component of the TFIID complex, and Inr-binding protein, respectively. TBP or Inr-binding protein plays a central role in recruiting basal transcription factors and RNA polymerase II forming a preinitiation complex (PIC). However, there are a few reports that have demonstrated that an activator-binding site alone is sufficient for a minimal promoter activity in transfected cells. These include the glucocorticoid response element (34) and the Ets motif (35-37). The mechanism by which the 29-bp sequence functions as a minimal promoter is not known. It is possible that USF bound to the E box stabilized the binding of TFIID to a cryptic TATA element through protein-protein interactions. Alternatively or additionally, USF bound to the E box can recruit TFIID and/or other components of the basal transcription machinery to the promoter and facilitate the assembly of the PIC. Several lines of evidence support this possibility. First, it has been reported (38, 39) that TBP is capable of binding to a number of sequences that are completely unrelated to the consensus sequence of the TATA element. Second, USF has been shown to be able to interact with TFIID (12, 40), increase the rate or stability of TFIID binding (41), and stabilize formation of the PIC (42). Third, USF is also able to interact with other transcription factors, including TBP associated factor TAFII55 (43), transcriptional cofactor PC5 (44), and TFII-I (45, 46) which can bind to both the Inr and the E box.

Another interesting aspect of this study is the distinct roles of the E box and the GC box in the transcriptional activity of the 69-bp sequence. Although the presence of the E box was essential for the transcriptional activity due to the GC box, the presence of the GC box was not required for the transcriptional activity due to the E box. This observation implied that USF mainly played a role in transcriptional initiation while the Sp1 stimulated the transcription. This observation is somewhat surprising since it has been shown (47-49) before that Sp1-binding site could direct transcriptional initiation of several TATA-less promoters. Our finding that USF-binding site was more efficient than that of Sp1-binding site in mediating transcriptional initiation could be a general phenomenon or restricted to a specific context of a promoter or cells. Nevertheless, our finding is in agreement with the hypothesis (50) that USF1 may not directly be responsible for transactivation but functions via interactions with other proteins. In this regard, it is interesting to note that USF1 does not only bind to the E box but also to the pyrimidine-rich Inr element (45, 46). In addition, ectopically expressed USF1 could stimulate transcription initiation through Inr element (46). Therefore, it is likely that the E box functions similarly to that of the Inr element if it is located near the transcriptional start site. Our finding that both the 69-bp (Fig. 1B) and the 29-bp sequence (data not shown) mediated transcription in a position-dependent manner also supports a recruiting rather than an activating function of the USF bound to the E box. Indeed, USF-binding site near the transcriptional start site has been found being essential for the basal promoter activity of some TATA-less promoters (15, 18, 19).

The functional significance of the role of the E box in TC II transcription is not known. Although the binding of USF to the E box may not affect the basal transcription of TC II in vivo due to its distal location from the start sites of the TC II gene, the potential binding in vivo of other members of bHLH/LZ family cannot be ruled out. Some of these could include Myc/Max, Mad/Max, and Max/Max. Elevation of plasma TC II levels is noted in a variety of cancers (51), and our hypothesis at the present time is that the E box by binding to Myc family may increase the transcription of the TC II gene which in turn will result in increased secretion of TC II to the circulation. Future studies with ectopic expression of the proteins belonging to the Myc family will address the role of these transcription factors in the in vivo transcription of the TC II gene.

In summary, we have identified a 69-bp DNA fragment containing a GC box and E box that is able to mediate transcription in vivo in an orientation-independent manner. The transcriptional activity was due to interplay between Sp1/Sp3 and USF1/USF2 that bind to these two cis-elements. Moreover, 29-bp fragment localized to the 3'-end of the 69-bp fragment containing only the E box was sufficient by itself to drive transcription, implying that USF bound to an E box can initiate and activate transcription in the absence of TATA box or other known Inr elements.