Downstream promoter sequences facilitate the formation of a specific transcription factor IID-promoter complex topology required for efficient transcription from the megalin/low density lipoprotein receptor-related protein 2 promoter.

Megalin/low density lipoprotein receptor-related protein 2 (LRP-2) is an endocytic receptor expressed in highly specialized cell types such as parathyroid cells and epithelia of the kidney. Previous experiments identified a nonconsensus TATA element, with the sequence TAGAAAA, as crucial for accurate and efficient transcription from the LRP-2 promoter. Here we show that, in addition to the TAGA element, promoter sequences downstream of the transcription start site contribute significantly to transcription both in vitro and in transfected cells. Deletion and point mutational analyses reveal that the promoter region located between positions +5 and +11 (sequence TTTTGGC) is of particular importance. Complementation experiments in nuclear extracts lacking transcription factor IID (TFIID) activity show that TATA-binding protein-associated factors of TFIID are essential for the function of LRP-2 downstream promoter sequences. Interestingly, DNase I footprinting studies show that the downstream region between positions +5 and +11 does not significantly affect overall TFIID affinity to the promoter but that it profoundly affects the topology of the TFIID x promoter complex not only downstream of the transcription start site, but in particular in the TATA box region. Our observations suggest a model for a novel downstream sequence function, in which TATA-binding protein-associated factor-promoter interactions downstream of the transcription start site modulate TFIID-DNA interactions in the TATA box region.

Megalin/low density lipoprotein receptor-related protein 2 (LRP-2) is an endocytic receptor expressed in highly specialized cell types such as parathyroid cells and epithelia of the kidney. Previous experiments identified a nonconsensus TATA element, with the sequence TAGAAAA, as crucial for accurate and efficient transcription from the LRP-2 promoter. Here we show that, in addition to the TAGA element, promoter sequences downstream of the transcription start site contribute significantly to transcription both in vitro and in transfected cells. Deletion and point mutational analyses reveal that the promoter region located between positions ؉5 and ؉11 (sequence TTTTGGC) is of particular importance. Complementation experiments in nuclear extracts lacking transcription factor IID (TFIID) activity show that TATA-binding protein-associated factors of TFIID are essential for the function of LRP-2 downstream promoter sequences. Interestingly, DNase I footprinting studies show that the downstream region between positions ؉5 and ؉11 does not significantly affect overall TFIID affinity to the promoter but that it profoundly affects the topology of the TFIID⅐promoter complex not only downstream of the transcription start site, but in particular in the TATA box region. Our observations suggest a model for a novel downstream sequence function, in which TATA-binding protein-associated factor-promoter interactions downstream of the transcription start site modulate TFIID-DNA interactions in the TATA box region.
The process of transcription underlies strong regulatory mechanisms, restricting expression of most genes to certain cell types or developmental stages. The establishment of cell-free systems, some 20 years ago, has permitted biochemical fractionation and purification of a set of six general transcription factors, denoted TFIIA, 1 -B, -D, -E, -F, and -H, that, in addition to RNA polymerase II, are minimally required for low (basal) levels of accurate transcription initiation in vitro. The activity of the general transcription machinery in vivo is regulated by gene-and cell type-specific transcription activators and repressors as well as various cofactors that are thought to function as mediators between regulatory proteins and individual components of the general transcription machinery (reviewed in Refs. 1

and 2).
A key step in the formation of functional transcription initiation complexes is the recognition of promoter sequences by components of the general transcription machinery. Indeed, it has become evident that the core promoter sequence context has a significant influence on both the overall efficiency of gene transcription and the ability of individual genes to respond to various transcription activators (1,3). Functional core promoter sequence elements identified to date include the TATA box, the initiator (Inr), the downstream promoter element (DPE), and the TFIIB recognition element. The best characterized core promoter elements are the TATA box, an A/T-rich sequence located about 25-30 nucleotides upstream of the transcription start site that is recognized by the TATA-binding subunit of TFIID (reviewed in Ref. 4), and the Inr element, a pyrimidine-rich sequence with the consensus YYA ϩ1 N(T/A)YY at the start site of transcription (5,6). Both TATA and Inr elements are sufficient to independently direct accurate transcription initiation in vitro but can also function in concert in a synergistic manner (see Ref. 7 and references therein). The DPE was recently identified as a sequence element in many TATA-less, Inr-containing promoters, located some 25-30 nucleotides downstream of the transcription start site with the consensus (A/G)G(A/T)CGTG (8,9). Finally, the TFIIB recognition element was identified, by binding site selection, as a GC-rich sequence located immediately upstream of the TATA box, which can enhance both binding of TFIIB to the TBP⅐DNA complex and promoter activity (10).
While the functions of the TATA box and the TFIIB recognition element sequences correlate well with their relative affinities to their cognate binding proteins TBP and TFIIB (10,11), studies in cell-free metazoan systems (7,12,13) and studies in yeast cells (14,15) have established a general requirement for TAF II s in the function of promoter regions downstream of the TATA element as well as for the transcription of TATA-less genes (7) where the TATA box-binding activity of TBP is largely dispensable (16). More specifically, functional studies with par-tially reconstituted recombinant Drosophila TFIID complexes have shown a minimal requirement of TBP, TAF II 250, and TAF II 150 for promoter-specific downstream sequence function (13). DNase I footprinting studies had demonstrated earlier that the human TFIID multiprotein complex, but not TBP alone, can make contact with promoter sequences downstream of the TATA element and the transcription start site of certain core promoter sequences (17)(18)(19). More recently, UV crosslinking experiments have examined the relative disposition of DNA and TFIID subunits within a human TFIID⅐promoter complex and have identified several human and Drosophila TFIID subunits in close proximity of promoter DNA. These studies suggest that Drosophila and human TAF II 250 and TAF II 150 might directly interact with the Inr element and promoter regions downstream of certain promoters (13,20,21) and that the histone-like Drosophila TFIID subunits TAF II 60 and TAF II 40 may contact the DPE (9). Based on these observations, it has been proposed that downstream promoter elements function in part by increasing TFIID⅐promoter complex formation and/or stability through direct interactions with TAF II s. Furthermore, dependent on the presence of TFIIA, TAF II interactions with downstream promoter regions may be concomitant with the formation of a stereospecific TFIID⅐promoter complex (21), consistent with a general requirement for TFIIA in core promoter-and TAF II -dependent transcription (22)(23)(24). Finally, a recent study has identified novel TAF II -and initiator-dependent cofactors (TICs) that are, in addition to TFIIA and TAF II s, required for core promoterspecific functions of TFIID through the initiator region of TATA-containing and TATA-less promoters (24). Taken together, these observations suggest that quantitative or qualitative core promoter sequence-dependent effects on TFIID promoter binding may constitute one important step for the functional readout of downstream core promoter sequence elements.
We previously identified a weak, atypical TATA box (TA-GAAAA) as necessary for transcription of the human megalin/ low density lipoprotein receptor-related protein 2 (LRP-2) gene promoter (25). Here, we demonstrate that efficient LRP-2 transcription additionally requires promoter sequences downstream of the transcription start site. Deletion and point mutation analyses indicate that stimulatory LRP-2 downstream promoter functions are conferred to a large extent by core promoter sequences located between positions ϩ5 and ϩ11, with no apparent sequence similarities to the Inr and DPE consensus sequences. DNase I footprinting experiments reveal that the presence of this sequence does not significantly affect the overall affinity of TFIID binding. Instead, TFIID interactions with the ϩ5 to ϩ11 region induce dramatic qualitative changes in TFIID interactions in the LRP-2 TATA box region, suggesting the formation of a conformationally distinct TFIID⅐promoter complex.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-A previously described fragment of the human LRP-2 promoter, containing sequences between positions Ϫ120 and ϩ52, cloned into the pGL-3 basic vector (25), was used as polymerase chain reaction template for construction of the deletion and point mutants outlined in Fig. 1A. Deletion of the LRP-2 promoter in the Ϫ120/ϩ11 and Ϫ120/ϩ4 constructs replaced the wild type LRP-2 sequence ϩ3 GCTTTTGGCCACTAGGAGCTGGCGGAGG ϩ30 with ϩ3 GCT-TTTGGCGCTAGCCCGGGCTCGAGAT ϩ30 and ϩ3 GCGCTAGCCCG-GGCTCGAGATCTGCGAT ϩ30 , respectively (vector sequences in boldface type; ϩ3 and ϩ30 refers to nucleotide positions relative to the transcription start site). To generate the Ϫ120/ϩ52(mut4 -10) construct, the wild type sequence ϩ3 GCTTTTGGC ϩ11 was changed to GTC-TCGGTC, while for construction of the Ϫ120/ϩ11(mut5-8), the same wild type sequence was changed to GCGCTAGGC. For analysis of promoter activity in vivo by RNase protection assay, a fragment con-taining the SV40 enhancer was inserted downstream of the luciferase gene in some of the reporters, as outlined in the text (see Fig. 1D). All constructs were verified by DNA sequencing.
Transfections and Reporter Gene Assays-JEG-3 cells were maintained and transfected as described previously (25). For luminometric determination of promoter activity, cells were transfected in 60-mm plates with 5 g of luciferase reporter and 1 g of CMV-LacZ in 10 g of total DNA. Cells were harvested 36 h post-transfection, and luciferase and ␤-galactosidase activities were determined luminometrically. For analysis of RNA, cells were transfected in 90-mm plates, with 10 g of luciferase reporter plasmid/SV40, 100 ng of OVEC-REF (28) and calf thymus DNA to 18 g total per plate. RNA was isolated 36 h after transfection, by the Nonidet P-40 lysis method (29). RNase protection analysis was performed as described (30), utilizing probes specific for the RNAs derived from the different mutant promoters. Products were quantified by PhosphorImager analysis. Numbers were normalized for differences in product content of the radiolabeled ribonucleotide ([ 32 P]UTP) and for differences in transfection efficiency, as determined by the signal obtained from the co-transfected OVEC-REF.
In Vitro Transcription and DNase I Footprinting-In vitro transcription experiments in untreated HeLa nuclear extracts were performed as described previously (7). Products were quantified by primer extension using the GL-2 sequencing primer (Promega). DNase I footprinting was done essentially as described (21). Templates containing the respective LRP-2 promoter inserts and flanking vector sequences, were radiolabeled on the noncoding strand, using polynucleotide kinase. Similar molar amounts (2.4 fmol/15 l) of different templates were used in the binding reactions. Products were analyzed on 6% (primer extension) or 8% (footprinting) sequencing gels and quantified by PhosphorImager analysis, and results were documented by autoradiography.

Downstream Core Promoter Sequences Contribute to Human
LRP-2 Gene Activity-We have shown previously that LRP-2 promoter activity is dependent on a weak nonconsensus TATA element with the sequence TAGAAAA located 30 bp upstream of the transcription initiation site (25). Earlier studies demonstrated that introduction of a G⅐C base pair at position 3 of the TATAAAA consensus sequence reduces TFIID promoter activity 50-fold (11). We therefore reasoned that additional core promoter sequences downstream of the TATA element may contribute to the LRP-2 promoter activity. Since the LRP-2 promoter appears to lack an initiator element (consensus YYA ϩ1 N(T/A)YY; Ref. 6), we examined the functional contribution of LRP-2 promoter sequences downstream of the transcription start site. Two 3Ј-deletion LRP-2 promoter mutants, encompassing nucleotides Ϫ120 to ϩ11 and Ϫ120 to ϩ4 (Fig. 1A), were constructed, and their transcription activities were compared with that of the full-length promoter (Ϫ120/ϩ52; Ref. 25) in vitro and in transfected cells.
As shown in Fig. 1B, RNA polymerase II-dependent LRP-2 transcription (compare lanes 1 and 2) was moderately reduced by deletion of the promoter region between ϩ11 and ϩ52 (com- pare lanes 1 and 3). Further deletion of sequences to ϩ4, however, resulted in a dramatic reduction of LRP-2 promoter activity, almost to background level (Fig. 1B, compare lanes 3  and 4). Importantly, introduction of clustered point mutations between ϩ5 and ϩ11 in constructs carrying LRP-2 promoter sequences either from Ϫ120 to ϩ52 (yielding Ϫ120/ϩ52 mut(4 -10)) or from Ϫ120/ϩ11 (yielding Ϫ120/ϩ11 mut(5-8); Fig. 1A) also strongly reduced LRP-2 promoter activity (Fig. 1B, compare lane 1 with lane 5, and compare lane 3 with lane 6). These observations demonstrate that core promoter sequences downstream of the transcription start site significantly contribute to efficient LRP-2 transcription in HeLa nuclear extracts and that a major proportion of LRP-2 downstream promoter function is conferred by the DNA sequence 5Ј-TTTTGGC-3Ј located between ϩ5 and ϩ11.
To study the function of LRP-2 downstream promoter regions in living cells, we co-transfected JEG-3 cells, which express LRP-2, with reporter plasmids containing the LRP-2 promoter variants fused to a luciferase reporter gene (Fig. 1A) and a ␤-galactosidase reference plasmid. As shown in Fig. 1C, deletion of LRP-2 promoter sequences from ϩ11 to ϩ52 reduced luciferase activities in the respective cell extracts by 50%, while further deletion of sequences from ϩ11 to ϩ4 resulted in an additional 2-3-fold reduction of LRP-2 promoter activity. Importantly, and in agreement with our in vitro data, clustered point mutation of promoter sequences between ϩ5 and ϩ11 affected LRP-2 transcription to the same extent as the deletion of promoter sequences between ϩ5 and ϩ11 (Fig. 1C). We confirmed that changes in luciferase activity in these assays FIG. 1. Downstream core promoter sequences contribute to human LRP-2 gene activity. A, nucleotide sequence of the human LRP-2 promoter between positions Ϫ40 and ϩ15 (top) and schematic description of the constructs used in this study. The TAGAAAA element is boxed, and the region between ϩ5 and ϩ11 studied extensively in this paper is underlined. The position of the start site of transcription (25) is indicated (arrow). B, in vitro transcription assays were performed with the indicated constructs, using 90 g of HeLa cell nuclear extract protein per reaction. Products were analyzed by primer extension. Relative transcription levels determined by PhosphorImager analysis (Molecular Dynamics, Inc., Sunnyvale, CA) are shown. C, transient expression of LRP-2 promoter variants in JEG-3 cells co-transfected with a CMV-LacZ reference plasmid. Cells were transfected in triplicate, and the mean relative luciferase activities normalized for differences in ␤-galactosidase activity are shown. The relative activity for each construct differed by less than 20% between experiments. D, quantitative RNase protection analysis of RNA prepared from JEG-3 cells co-transfected in duplicate with the indicated LRP-2 reporter constructs and the reference plasmid, OVEC-REF. Lane 4 represents a reaction run with RNA from nontransfected cells. The relative transcription levels, normalized for differences in product content of radiolabeled ribonucleotide ([ 32 P]UTP) and reference gene signal, are shown below the autoradiogram. indeed reflected changes in intracellular luciferase mRNA levels by performing RNase protection assays (Fig. 1D) on total RNA from cells co-transfected with derivatives of our luciferase reporter gene constructs carrying the SV40 enhancer upstream of the LRP-2 promoter and the OVEC-REF plasmid as a reference (28).
Taken together, these findings demonstrate a requirement of core promoter sequences downstream of the transcription start site for efficient LRP-2 transcription in vitro and in transfected cells. Our data also suggest that a major proportion of LRP-2 downstream promoter function is conferred by the DNA sequence 5Ј-TTTTGGC-3Ј located between positions ϩ5 and ϩ11. This sequence appears not to be related to a functional initiator element (consensus YYA ϩ1 N(T/A)YY; Ref. 6) or to the recently described DPE in Drosophila (consensus (A/G)G(A/T)CGT; Refs. 8 and 9) and may therefore contain a novel downstream core promoter element.
The LRP-2 Downstream Sequence Function Is TAF II -dependent-Work from a number of laboratories has clearly established that TAF II s perform critical functions in the binding and the functional readout of core promoter elements distinct from the TATA element (reviewed in Refs. 1, 3, and 4). To test for a role of TAF II s in LRP-2 downstream promoter sequence function, we examined the function of LRP-2 downstream promoter sequences between ϩ5 and ϩ52 on transcription in a HeLa nuclear extract immunodepleted of TFIID activity (NE[⌬D]; Ref. 27). As shown in Fig. 2B, transcription from the LRP-2 promoter in this extract was completely dependent on the addition of either highly purified recombinant human 6His:TBP or equivalent amounts (TBP content) of immunoaffinity-purified f:TFIID ( Fig. 2A; Refs. 19 and 27). In the presence of f:TFIID, the relative levels of transcription from the parental Ϫ120/ϩ52 constructs and the two deletion mutants in NE [⌬D] were comparable with those observed in untreated extract (compare lanes 3, 6, and 9 of Fig. 2B with lanes 1, 3, and 4 of Fig. 1B). Thus, the addition of highly purified f:TFIID ( Fig. 2A) was sufficient to restore LRP-2 downstream sequence functions in NE [⌬D]. Importantly, transcription levels from the LRP-2 promoter containing downstream promoter sequences up to ϩ52 was significantly higher (Ͼ5-fold) with f:TFIID than with similar amounts of human 6His:TBP (Fig. 2B, compare lane 2 with lane 3). This was also true for the Ϫ120/ϩ11 construct (lanes 5 and 6), although the difference in absolute transcription levels was somewhat lower (3.4-fold). Despite the very weak absolute transcription levels, transcription from the Ϫ120/ϩ4 template was reproducibly higher with 6His:TBP than with f:TFIID (lanes 8 and 9).
Taken together, these results demonstrate that LRP-2 downstream sequence function is TAF II -dependent. They also indicate that TAF II s may actually inhibit TBP function in the absence of LRP-2 downstream promoter sequences, consistent with earlier observations with other core promoters (13,27,31). Similar results were obtained in analogous experiments using heat-treated HeLa nuclear extracts that lack TFIID activity (Ref. 17; data not shown).

LRP-2 Downstream Sequences Facilitate Promoter Binding by Components of TFIID in the Presence of the Weak Noncon-
sensus TATA Element-To further investigate the role(s) of TAF II s in LRP-2 downstream promoter sequence function, we carried out DNase I footprinting experiments using radiolabeled DNA fragments containing various LRP-2 promoter variants, immunoaffinity-purified f:TFIID, and partially purified human TFIIA (21). Binding of f:TFIID to a DNA fragment containing LRP-2 promoter sequences from Ϫ120 to ϩ52 was relatively weak, presumably due to the low affinity TATA box sequence present in the LRP-2 promoter (Ref. 11; see below).
However, detectable DNase I protection extended over a broad promoter region from about position Ϫ40 to ϩ30 (Fig. 3, lanes  1-4). Importantly, f:TFIID binding induced strong DNase I hypersensitivity centered around positions Ϫ50 and Ϫ60 upstream of the TAGA element, as well as within the LRP-2 downstream promoter region at positions ϩ7 and ϩ8.
Removal of LRP-2 downstream sequences from ϩ11 to ϩ52 resulted in a modest reproducible reduction of overall f:TFIID binding to the LRP-2 promoter but did not affect DNase I hypersensitivity upstream from the TATA element and downstream from the transcription start site (Fig. 3, compare lanes  1-4 with lanes 5-8), in agreement with the modest, yet reproducible impairment of transcription observed in vitro and in vivo (Fig. 1, B, C, and D). Importantly, and in contrast to our observations with the Ϫ120/ϩ52 and Ϫ120/ϩ11 constructs, further deletion of LRP-2 promoter sequences to ϩ4 did not significantly reduce overall f:TFIID binding but instead completely abolished the induction of DNase I hypersensitivity upstream from the TATA element and downstream from the transcription start site (Fig. 3, compare lanes 9 -12 with lanes  5-8). These results suggest that LRP-2 promoter sequences from ϩ5 to ϩ11 are major determinants of f:TFIID⅐DNA complex formation, consistent with their importance for efficient LRP-2 transcription in vitro and in transfected cells (Fig. 1, B,  C, and D).
Finally, we assessed the contribution of the nonconsensus LRP-2 TATA element in f:TFIID promoter binding by performing DNase I footprinting assays with a promoter construct containing LRP-2 sequences from Ϫ10 to ϩ52, lacking the TAGA element. Surprisingly, we were able to detect weak f:TFIID binding, protecting a region from DNase I digestion downstream of the transcription start site and to around position Ϫ40 (Fig. 3, lanes 13-15). Thus, although the TAGA element at position Ϫ30 is required for detectable LRP-2 transcription in vitro (25), LRP-2 downstream promoter sequences between positions Ϫ10 and ϩ52 appear to be sufficient for low affinity f:TFIID binding.
Taken together, these observations demonstrate that DNA sequences downstream of the transcription start site contribute to efficient f:TFIID binding. Interestingly, the presence of the downstream region between ϩ5 and ϩ11 did not significantly increase the overall affinity of f:TFIID binding to the LRP-2 promoter (Fig. 3, lanes 5-12) but rather induced strong DNase I hypersensitivity upstream of the TATA element and downstream of the LRP-2 transcription start site. This indicated to us that this particular part of the LRP-2 downstream region may affect the overall topology of the f:TFIID⅐LRP-2 promoter complex (see below).
We attempted unsuccessfully to identify the component(s) of TFIID that contact the LRP-2 downstream promoter region by a photocross-linking method (data not shown) using promoter constructs modified with the thymidine derivative 5-[N-(pazidobenzoyl)-3-aminoallyl]-deoxyuridine triphosphate (32). The lack of specifically cross-linked products may be explained by the fact that the method only probes the vicinity of the DNA major groove and hence would fail to detect other interactions. Alternatively, the bulky side chain of the 5-[N-(p-azidobenzoyl)-3-aminoallyl]-deoxyuridine monophosphate residues may have interfered with the respective TFIID-promoter interactions.
Promoter Regions Downstream from the Transcription Start Site Mediate Conformational Changes within the TFIID⅐ Promoter Complex at Promoter Regions Upstream from the TATA Element and at the Transcription Start Site-In order to distinguish qualitative and quantitative effects of LRP-2 downstream promoter sequences on TFIID binding, we used LRP-2 promoter constructs engineered to contain a consensus TATA element, designated Ϫ120/ϩ52(G3T) and Ϫ120/ϩ4(G3T), respectively ("G3T" for TATAAAA; Fig. 1A). Consistent with previous observations (25), conversion of the TAGAAAA sequence to a consensus TATA box increased LRP-2 promoter activity more than 10-fold (Fig. 4A, compare lanes 1 and 3). Interestingly, deletion of promoter sequences downstream from the transcription start site reduced LRP-2 transcription from promoter constructs carrying a consensus TATA element and promoter constructs with the wild type TAGA element to an approximately similar extent. This result suggests that LRP-2 downstream promoter functions are independent of a particular TATA element sequence. Furthermore, absolute transcription levels from LRP-2 promoter constructs that contained a consensus TATA box but lacked LRP-2 downstream sequences were comparable with absolute transcription levels from constructs containing wild type LRP-2 promoter sequences from Ϫ120 to ϩ52 (Fig. 4A, compare lanes 1 and 4). Thus, a consensus TATA sequence can substitute for promoter sequences downstream of ϩ4 to mediate wild type levels of LRP-2 transcription.
Next, we performed DNase I footprinting experiments to examine the effects of LRP-2 downstream sequences on f:TFIID promoter binding in the presence of a consensus TATA box. As expected, conversion of the TAGA element into a consensus TATA box increased f:TFIID binding to the LRP-2 promoter, leading to increased protection of the TATA box region and downstream promoter sequences (Fig. 4B, compare lanes 1-4  with lanes 5-8). However, in contrast to the results obtained with promoter templates containing the wild type LRP-2 TATA sequence TAGAAAA, deletion of LRP-2 downstream sequences between ϩ5 and ϩ52 had no significant quantitative effect on f:TFIID promoter binding to DNA templates containing a consensus TATA element (Fig. 4B, compare lanes 5-8 with lanes  9 -12; Fig. 3). Most importantly, while the overall levels of protection in the TATA box region remained essentially unchanged, the absence of LRP-2 promoter sequences down- stream of ϩ4 resulted in a complete loss of the DNase I hypersensitivity observed with the full-length promoter constructs in the regions upstream from the TATA box and downstream from the transcription start site. This observation implies that TFIID interactions with LRP-2 downstream promoter regions between positions ϩ5 and ϩ52 not only contribute to the overall TFIID affinity (detectable only in the context of a weak TATA sequence; see Fig. 3) but in addition result in a significant reorganization of the TFIID⅐DNA nucleoprotein complex topology. Of particular interest is that interactions of TFIID with LRP-2 promoter regions downstream of the transcription start site, largely mediated by DNA sequences located between ϩ5 and ϩ11 (Fig. 3), strongly affect TFIID-DNA interactions at the TATA region, 30 base pairs upstream of the transcription start site. These qualitative effects of LRP-2 downstream se-quences on TFIID binding correlate well with increased transcription in HeLa nuclear extract (Figs. 1B, 2B, and 4A). Therefore, our data suggest that LRP-2 promoter sequences downstream of the transcription start site, in particular the downstream region located between ϩ5 and ϩ11, contribute to LRP-2 transcription by facilitating the formation of a specific TFIID⅐promoter complex topology that is required for efficient preinitiation complex formation.

DISCUSSION
TAF II -dependent Transcription through Downstream Sequences in the LRP-2 Gene-Our previous experiments showed that a TAGAAAA sequence in the Ϫ30 region was required for human LRP-2 promoter activity (25). However, earlier reports suggested that a TAGA element would not be sufficient in mediating efficient TFIID binding and transcription (11). We therefore looked for additional core promoter sequences that would contribute to high levels of LRP-2 transcription.
The LRP-2 transcription start site sequence is rich in purines (Fig. 1A) and shows no similarity to the Inr element (consensus YYA ϩ1 N(T/A)YY; Ref. 6). Furthermore, mutation of a LRP-2 promoter sequence centered around ϩ29 with homology to the DPE consensus sequence (8) had no detectable effect on TFIID binding and transcription (data not shown). Instead, our deletion and point mutation analyses suggest that a promoter region with the sequence 5Ј-TTTTGGC-3Ј located between positions ϩ5 and ϩ11 is of particular importance for efficient LRP-2 transcription. We were unable to detect significant homologies with this sequence in any other TATA-containing and initiatorless promoter in the Eukaryotic Promoter Database (33), including class II promoters for which downstream interactions with human TFIID have been demonstrated (17,18). Based on these results, we speculate that the LRP-2 downstream region between ϩ5 and ϩ11 may constitute a novel class II core promoter sequence element.
Using HeLa cell nuclear extracts immunodepleted of TBP and TFIID-specific TAF II s as a cell-free transcription system, we compared the effect of the LRP-2 downstream sequences in TBP-and TFIID-directed transcription. In agreement with earlier studies on core promoter-specific transcription (7,12,13), we find that LRP-2 downstream sequence function is dependent on TAF II s within the TFIID complex. We also noted that, in the absence of the LRP-2 downstream sequences, TBP mediated higher absolute levels of transcription than TFIID, whereas in the presence of the LRP-2 downstream sequences transcription levels with TFIID were higher than with TBP. This may indicate that TAF II s actually repress TBP (TFIID) function in the absence of LRP-2 downstream promoter functions, similar to observations with other core promoter sequences (13,18,27,31). Thus, the stimulatory effect of the LRP-2 downstream region may reflect the relief of TAF II -mediated repression as well as a net stimulation of TBP (TFIID) function (Ref. 34; see below).
Effects of LRP-2 Downstream Promoter Regions on TFIID Binding-Previous studies have demonstrated that downstream promoter sequences can exert quantitative (overall binding affinity) as well as qualitative (overall topology) effects on TFIID-promoter interactions (reviewed in Refs. 1 and 4). We find that the LRP-2 downstream region stimulates TFIID binding to the LRP-2 promoter in the presence of TFIIA only very weakly. Instead, DNase I-hypersensitive sites indicated dramatic differences in the overall conformation of the TFIIA⅐TFIID⅐LRP-2 promoter complex in the absence and presence of the LRP-2 downstream region. Importantly, changes in TFIID⅐promoter complex topology occurred not only within the ϩ5 to ϩ11 region itself but were particularly evident upstream of the LRP-2 TATA box. These effects are TAF II -dependent, since no DNase I hypersensitivity upstream of the TATA box region was observed when bacterially expressed human TBP was used (data not shown). The lack of DNase I hypersensitivity upstream from the LRP-2 TATA region may indicate that TBP-TATA interactions in the absence of LRP-2 downstream sequences are compromised by the presence of TAF II s within the TFIID complex. Thus, our results provide, for the first time, evidence that TAF II interactions with promoter sequences downstream the transcription start site can in turn exert qualitative effects on TFIID-promoter interactions around and upstream of the TATA element.
A Potential Model for LRP-2 Downstream Sequence Function-Earlier work has established a correlation between activator-induced (and TFIIA-dependent) isomerization of a hu-man TFIIA⅐TFIID⅐promoter complex characterized by the induction of TFIID downstream interactions, facilitated recruitment of the remaining general transcription factors, and stimulation of transcription (35)(36)(37). Thus, core promoter-dependent formation of a stereospecific TFIID⅐promoter complex may similarly facilitate functional preinitiation complex formation, possibly in conjunction with additional soluble core promoter sequence-specific cofactors, such as the recently identified TICs (24). However, depending on the core promoter sequence context, TAF II s have also been reported to repress TBP function (13,18,27,31). In particular, the N terminus of Drosophila TAF II 230, the human TAF II 250 homologue, inhibits promoter binding of TBP (38) through interactions of an N-terminal subdomain I with the TBP DNA-binding surface via structural mimicry of TBP-TATA box interactions (39). Remarkably, a second conserved N-terminal subdomain II, that contributes to TAF II 230-mediated repression of TBP function, interacts with residues on the convex surface of TBP that are crucial for interaction with TFIIA (40). Indeed, TFIIA seems to be generally required for TAF II -and core promoterspecific transcription (22)(23)(24) and derepression of TAF II 250mediated inhibition of human TFIID (41). In addition, the C-terminal portion of Drosophila TAF II 230/human TAF II 250 FIG. 5. A potential model for LRP-2 downstream sequence function. A, binding of human TFIID to the LRP-2 promoter requires (partial) relief from TAF II -mediated TFIID autoinhibition, e.g. displacement of the N-terminal domain I (NЈI) of TAF II 250 from the DNAbinding surface of TBP (34,39). B, in the absence of the LRP-2 downstream promoter regions, additional inhibitory interactions within the TFIID complex, e.g. between TBP and TAF II 250/230 N-terminal domain II (NЈII; Ref. 40), remain unaffected. C, interactions between TAF II s within TFIID and LRP-2 downstream promoter sequences reverse remaining autoinhibitory TFIID functions via conformational changes within the TFIID⅐DNA complex, allowing efficient preinitiation complex formation and subsequent transcription initiation. The identity of TFIID components that contact the LRP-2 downstream promoter region is not known. appears to be involved in the control of the inhibitory N-terminal domain (38). Keeping in mind that Drosophila TAF II 230 and its human homologue TAF II 250 have also been implicated in direct TFIID-promoter interactions at and downstream of the transcription start site (13,21), it is conceivable that interactions of the C-terminal portion of Drosophila TAF II 230/human TAF II 250 with promoter DNA contribute to relief TFIID autoinhibition by the N-terminal domains in conjunction with TFIIA. Based on our data and the considerations mentioned above, we favor a model in which LRP-2 core promoter sequences located between positions ϩ5 and ϩ11 stimulate LRP-2 transcription by facilitating relief from TFIID autoinhibition through TAF II s (Fig. 5).
Potential Role(s) of Downstream Promoter Regions in LRP-2 Gene Regulation-We found that an LRP-2 promoter template containing a consensus TATA element but lacking sequences downstream of ϩ4, directed similar levels of transcription as the natural full-length promoter containing the weak TAGA element. This indicates that one role of the downstream promoter region of the LRP-2 gene may be to complement the weak TAGA element, to reach a certain level of basal promoter strength. Also, we note obvious parallels between the LRP-2 gene and the human gene for glial fibrillary acidic protein, whose expression also is highly tissue-specific (see Ref. 42 and references therein). Both the LRP-2 and glial fibrillary acidic protein promoters display stimulation of TFIID binding and transcription by downstream promoter regions in the context of very weak TATA sequences (see Refs. 18 and 25; this paper). Thus, such a promoter architecture may be preferable for cell type-specific gene expression.
Given that activators may stimulate transcription by affecting the ability of TAF II s to interact with promoter regions at and downstream of the transcription start site, a dynamic functional interplay of activator-independent TAF II -promoter interactions and TAF II -mediated autoinhibition of TFIID may provide for additional levels of transcription regulation. It may also contribute to core promoter-specific activity of various activators (43)(44)(45)(46)(47)(48), encouraging speculation that the unique structure of the LRP-2 core promoter may be a prerequisite for productive interplay with an as yet unidentified enhancer of this gene.