Transcriptional activation of the rice tungro bacilliform virus gene is critically dependent on an activator element located immediately upstream of the TATA box.

To investigate the transcriptional mechanisms of rice tungro bacilliform virus, we have systematically analyzed an activator element located immediately upstream of the TATA box in the rice tungro bacilliform virus promoter and its cognate trans-acting factors. Using electrophoretic mobility shift assays, we showed that rice nuclear proteins bind to the activator element, forming multiple specific DNA-protein complexes via protein-protein interactions. Copper-phenanthroline footprinting and DNA methylation interference analysis indicated that multiple DNA-protein complexes share a common binding site located between positions -60 to -39, and the proteins contact the activator element in the major groove. DNA UV cross-linking assays further showed that two nuclear proteins (36 and 33 kDa), found in rice cell suspension and shoot nuclear extracts, and one (27 kDa), present in root nuclear extracts, bind to this activator element. In protoplasts derived from a rice (Oryza sativa) suspension culture, the activator element is a prerequisite for promoter activity and its function is critically dependent on its position relative to the TATA box. Thus, transcriptional activation may function via interactions with the basal transcriptional machinery, and we propose that this activation is mediated by protein-protein interactions in a position-dependent mechanism.

To investigate the transcriptional mechanisms of rice tungro bacilliform virus, we have systematically analyzed an activator element located immediately upstream of the TATA box in the rice tungro bacilliform virus promoter and its cognate trans-acting factors. Using electrophoretic mobility shift assays, we showed that rice nuclear proteins bind to the activator element, forming multiple specific DNA-protein complexes via protein-protein interactions. Copper-phenanthroline footprinting and DNA methylation interference analysis indicated that multiple DNA-protein complexes share a common binding site located between positions ؊60 to ؊39, and the proteins contact the activator element in the major groove. DNA UV cross-linking assays further showed that two nuclear proteins (36 and 33 kDa), found in rice cell suspension and shoot nuclear extracts, and one (27 kDa), present in root nuclear extracts, bind to this activator element. In protoplasts derived from a rice (Oryza sativa) suspension culture, the activator element is a prerequisite for promoter activity and its function is critically dependent on its position relative to the TATA box. Thus, transcriptional activation may function via interactions with the basal transcriptional machinery, and we propose that this activation is mediated by protein-protein interactions in a position-dependent mechanism.
In eukaryotes, the transcription of protein-coding genes by RNA polymerase II is regulated via two distinct types of DNA sequences: core promoter elements, located near the transcription start site that are sufficient to direct the accurate initiation of transcription; and upstream promoter elements, which contain binding sites for sequence-specific transcriptional activator and/or repressor proteins (1,2). Transcriptional activators stimulate transcription by recruiting the RNA polymerase II machinery to a core promoter and/or stabilizing the transcription-initiation complex (3)(4)(5)(6). Activators have been proposed to directly or indirectly (through coactivators) interact with components of the basal transcription machinery in mammalian systems (3,7,8).
Although the basal transcriptional machinery, also referred to as the polymerase II transcription initiation complex, has not been studied in plants as extensively as in mammalian, Drosophila, and yeast systems, TATA-binding proteins (TBPs), 1 TFIID, TFIIA, and RNA polymerase II subunits have been isolated from plants (9 -12). Moreover, a number of plant trans-acting factors have been identified (13). Little is known in plants about the molecular mechanisms of transcriptional activation. It has been shown that TGA1a, a transcription activator interacting with the activation sequence-1 element in the CaMV 35 S promoter (14), increases the rate of preinitiation complex formation (15,16). The plant transcription factor GT-1 belongs to the class of trihelix DNA-binding proteins, and it binds to a promoter cis-element with the core DNA sequence 5Ј-GGTTAA found initially in light-regulated genes (17). Recently, it has been shown that Arabidopsis GT-1 can interact with and stabilize the TFIIA-TBP-TATA complex, suggesting that GT-1 may activate transcription through direct interaction with the transcriptional pre-initiation complex (18).
Rice tungro bacilliform virus (RTBV) is a plant pararetrovirus (19 -22) and belongs to the "RTBV-like" genus of the caulimoviridae family (23). RTBV has a circular, double-stranded DNA genome of 8 kilobase pairs from which the terminally redundant pregenomic transcript is produced (19,20,24). A single promoter from the virus genome was isolated, and phloem-specific activity was observed in transgenic rice plants (25), suggesting that the promoter can account for the tissue-specific localization of this virus. The RTBV promoter has been analyzed in both transformed rice plants (26 -28) and transfected rice protoplasts (29,30). Multiple upstream elements have been identified as being required for phloem-specific gene expression in the context of the Ϫ164 to ϩ45 promoter in transformed rice plants (26,27). Moreover, the full-length RTBV promoter including the first 250 nucleotides downstream of the transcription start site is active in a wider range of rice cell types than one where these sequences are lacking (28). A promoter element required for gene expression in the vascular tissue was located to sequences between Ϫ165 and Ϫ100 to which proteins from rice nuclear extracts bind (28). In transfected protoplasts, the activity of the RTBV promoter also required downstream promoter elements located in the region from ϩ1 to ϩ90 (29,30).
In the current study, we have investigated the cis-acting element located immediately upstream of the TATA box in considerably more detail. We have compared the binding activity of different types of nuclear proteins and characterized the nuclear factors that bind to this element by DNA UV crosslinking assays. Furthermore, we have analyzed its function by deletion and mutation analysis in transfected protoplasts. The results reveal this element to be a prerequisite for promoter activity. Its function is critically dependent on its position relative to the TATA box. We propose that activation of the RTBV promoter by this element is mediated by a position-dependent transcriptional activation mechanism.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-The basic plasmid used in this study, R-218.I-CAT (here called R-218), has been described previously (29,30). The 5Ј deletions and point mutations were made by PCR with appropriate synthetic oligonucleotides. A common 5Ј end oligonucleotide primer, designated P5, was located at vector sequences upstream of the RTBV promoter. A common 3Ј end oligonucleotide primer, designated P3, covered the first 20 nucleotides of the CAT ORF on the antisense strand.
To generate plasmids with 5Ј end deletions to Ϫ70 or Ϫ40, a DNA fragment was amplified from the R-218 template by PCR with a 5Ј end primer corresponding to position Ϫ70 or Ϫ40 flanked by an XbaI site and primer P3. The PCR products were cloned into the XbaI-XhoI site of R-218 to generate plasmids R-70 or R-40, respectively.
The internal deletion from Ϫ70 to Ϫ35 was constructed as follows: a DNA fragment covering the region from Ϫ218 to Ϫ70 was amplified by PCR with primer P5 and a primer corresponding to nucleotides from Ϫ85 to Ϫ70 in antisense. A second DNA fragment (from Ϫ35 to first 20 nucleotides of the CAT ORF) was amplified with a primer corresponding to nucleotides from Ϫ35 to Ϫ20 and primer P3. These two PCR products were then phosphorylated and digested with XbaI or XhoI, respectively. The resulting DNA fragments were ligated together into the XbaI-XhoI site of R-218 to yield R-218d.
To obtain eR-218d(ϩ) and eR-218d(Ϫ), double-stranded oligonucleotides corresponding to nucleotides from Ϫ70 to Ϫ35 flanked with an XbaI site were inserted into XbaI site of R-218. The resulting constructs were designated as eR-218d(ϩ) with the insertion in the sense orientation and eR-218d(Ϫ) in antisense orientation.
For point mutations, an overlap extension PCR strategy was used: two overlapping fragments, one amplified with primer P5 and a 3Ј end primer of the antisense strand with mutation of Gs directly contacted by nuclear proteins to Ts (see below), and the other with a 5Ј-primer corresponding to the sense strand with mutation of Gs directly contacted by nuclear proteins to Ts and primer P3, were used as templates to amplify a mutated promoter fragment using P5 and P3 primers. The resulting product was digested with XbaI and XhoI, and introduced between the XbaI and XhoI sites of R-218 to yield R-218m.
Plasmids (AATTG) 1 , (AATTG) 2 , (AATTG) 3 , and (AATTG) 4 were created as follows: an upstream fragment covering the region from Ϫ218 to Ϫ35 was amplified with primer P5 and a primer corresponding to the antisense strand from Ϫ50 to Ϫ35 flanked with a MunI site (CAATTG). A downstream fragment from Ϫ34 to the first 20 nucleotides of the CAT ORF was made by PCR with a 5Ј-primer corresponding to sequences from Ϫ34 to Ϫ20 flanked with a MunI site and primer P3. The resulting PCR products thus harbor AATTG. Downstream fragments bearing (AATTG) 2 , (AATTG) 3 , or (AATTG) 4 at the 5Ј end were produced as above. The upstream and downstream fragments were digested with MunI and ligated, followed by digestion with XbaI and XhoI. The resulting fragments were inserted between XbaI and XhoI sites of R-218.
A DNA fragment spanning positions Ϫ100 to Ϫ1 amplified by PCR was inserted into the HindIII site (filled in with Klenow polymerase) of pBluescript II KS(ϩ) to generate plasmids for preparation of DNA probes. The resulting plasmids were designated as pRP100(ϩ) with the 5Ј end of the sense strand adjacent to the ClaI site, and pRP100(Ϫ) in antisense orientation. To create plasmid mP100 for preparation of the DNA probe used in UV cross-linking assays, a PCR fragment covering positions Ϫ100 to Ϫ1 of the promoter, flanked by an XbaI site at the 5Ј end and HindIII at the 3Ј end, was inserted between the XbaI and HindIII sites of M13mp18.
PCR conditions were as follows: one cycle at 94°C for 3 min, 30 cycles at 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min, followed by one cycle at 72°C for 10 min. Pfu DNA polymerase (Stratagene) was used in all the PCR reactions. All deletions and mutations were verified by restriction digestions and DNA sequencing. Plasmids were isolated from Escherichia coli (strain DH5␣) using a Qiagen plasmid kit.
Preparation of Nuclear Extracts-Nuclear extracts from cell suspensions of Oryza sativa line Oc were prepared as described (30). Nuclear extracts from 2-week-old O. sativa plant shoots and roots were prepared essentially as described (31). All steps were performed at 4°C. Briefly, 100 g of fresh tissue of shoots or roots was homogenized in 400 ml of cold homogenization buffer (1 M 2-methyl-2,4-pentanediol, 10 mM PIPES/ KOH, pH 7.0, 10 mM MgCl 2 , 0.5% Triton X-100, 5 mM 2-mercaptoethanol, 0.8 mM phenylmethylsulfonyl fluoride). The homogenate was filtered sequentially through two layers of gauze, 60-m and then 40-m nylon mesh filters. The nuclei were pelleted by centrifugation at 3,000 ϫ g for 10 min and resuspended in 100 ml of nuclear wash buffer with Triton (0.5 M 2-methyl-2,4-pentanediol, 10 mM PIPES/KOH, pH 7.0, 10 mM MgCl 2 , 5 mM 2-mercaptoethanol, 0.8 mM phenylmethylsulfonyl fluoride, 0.5% Triton X-100). The solution was filtered through 10-m nylon mesh filters, and the nuclei were collected by centrifugation at 3000 ϫ g for 5 min. The nuclear pellet was resuspended in 50 ml of nuclear wash buffer without Triton. After centrifugation as above, the nuclei were resuspended in 20 ml of nuclear lysis buffer (110 mM KCl, 15 mM Hepes/KOH, pH 7.5, 5 mM MgCl 2 , 1 mM dithiothreitol, 5 g/ml antipain, and 5 g/ml leupeptin), followed by addition of three aliquots of 4 M ammonium sulfate (total volume of 2 ml) with gentle mixing. After shaking gently for 30 min, the solution was centrifuged to pellet the chromatin and particulate matter at 40,000 rpm for 45 min. Ammonium sulfate (0.30 g/ml) was then added to the supernatant with gentle stirring over a 30-min period, and the solution was left to stand for 30 min. The solution was then centrifuged at 10,000 rpm for 15 min to precipitate protein, and the pellet was resuspended gently in 0.5-1 ml of nuclear extract buffer (40 mM KCl, 25 mM Hepes/KOH, pH 7.5, 0.1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 5 g/ml antipain, 5 g/ml leupeptin), followed by dialysis against nuclear extract buffer (dithiothreitol was replaced by 5 mM 2-mercaptoethanol and protease inhibitors were omitted) three times for 1 h. The dialyzed solution was centrifuged at 10,000 rpm for 10 min, frozen in liquid nitrogen, and stored in aliquots at Ϫ80°C. Protein concentration was measured by the method of Bradford (Bio-Rad) with bovine serum albumin as the standard.
Electrophoretic Mobility Shift Assay (EMSA)-A DNA probe was 5Ј end-labeled with [␣-32 P]dCTP using the Klenow DNA polymerase with ClaI-digested pRP100(ϩ), followed by a second restriction digestion at the EcoRV site. The resulting probe was purified on a 5% native polyacrylamide gel. Binding reactions were carried out essentially as described (30), except that 15,000 cpm of labeled DNA (around 0.4 ng of DNA) and 10 -20 g of nuclear extract proteins were used for each reaction. 5 g of poly(dI-dC) (Amersham Pharmacia Biotech) was used as an unspecific competitor.
For treatments with sodium desoxycholate (DOC), DOC was added to the reaction mixture at the amounts indicated 20 min after addition of the labeled DNA probe, and the mixtures were incubated at room temperature for an additional 10 min. As a control, Nonidet P-40 was added to a final concentration of 1% to reverse the function of DOC as detergent.
Copper-Phenanthroline Footprinting Analysis-The same 5Ј end-labeled DNA probe was used as for EMSA. A 3Ј end-labeled DNA probe was prepared in the same way as the 5Ј end-labeled probe except using plasmid pRP100(Ϫ). The footprinting analysis was performed as described (32) with some modifications. Binding reactions were scaled up 10-fold as compared with the EMSA and fractionated by polyacrylamide gel electrophoresis as with EMSA. In situ digestion of the DNA by the nuclease activity of 1,10-phenanthroline-cuprous complex was allowed to proceed for 5-6 min at 25°C. DNA from free and bound fractions were visualized by autoradiography of the wet gel at 4°C overnight, then the DNA was eluted from the corresponding bands in 300 l of elution buffer (20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 0.4 M NaCl, 0.05% SDS) overnight at 37°C. The DNA was extracted with phenol-chloroform and purified using a Qiagen PCR purification kit. Equal Cerenkov counts of the bound and free DNA fractions were loaded on a 6% polyacrylamide sequencing gel. The gel was then fixed, dried, and exposed for autoradiography. Chemical sequencing reactions were carried out according to the Maxam and Gilbert method (33).
Methylation Interference Footprinting Analysis-The probes and the binding reactions were exactly the same as those used for copperphenanthroline footprinting analysis. The labeled probe was partially methylated with dimethyl sulfate (34). EMSAs were then carried out as described above with the methylated probe. Following native gel electrophoresis, the unbound probe and protein-DNA complexes were located by autoradiography, excised, and eluted from the gel overnight at 37°C in elution buffer (see above). Proteins were removed by extraction with phenol-chloroform, and the DNA was purified using a Qiagen PCR purification kit. The DNA was then cleaved in 1 M piperidine at 90°C for 30 min, lyophilized, dissolved in formamide gel-loading dye, and resolved in a 6% sequencing gel.
DNA UV Cross-linking Assays-Single-stranded mP100 DNA was prepared as described (33). DNA probe that has had its thymidine residues substituted by bromodeoxyuridine was uniformly labeled with [␣-32 P]dCTP by extension of the M13 universal primer as described (35), and then digested with XbaI and HindIII, followed by purification on a 5% native polyacrylamide gel. The binding reactions were carried out as described for EMSA and included 1 ϫ 10 5 cpm of the uniformly labeled DNA probe. Following the 20-min incubation period, the reaction mixtures were irradiated with UV light (312 nm) for 30 min in a UV Stratalinker 1800 (Stratagene) at 4°C. Samples were then digested with 10 units each of DNase I and micrococcal nuclease in the presence of 10 mM CaCl 2 for 30 min at 37°C. The resulting mixtures were resolved on a 14% SDS-polyacrylamide gel. Transient Expression Assays-Protoplasts derived from a rice Oc suspension culture were prepared and transfected by polyethylene glycol-mediated transfection as described previously (36,37). 10 g of plasmid DNA was used to transfect 0.6 ϫ 10 6 protoplasts, together with 3 g of a plasmid expressing ␤-glucuronidase (GUS) under the control of the CaMV 35 S promoter to serve as an internal standard for transfection efficiency. Measurement of CAT and GUS activities in protoplast extracts prepared after overnight incubation was carried out by using a CAT enzyme-linked immunosorbent assay kit (Roche Diagnostics Ltd) and as described elsewhere (38), respectively. CAT activity was normalized to the GUS activity of the same extract. Relative expression levels did not vary more than Ϯ10% with the given constructs in this study. All constructs were tested in at least three independent experiments with at least three independent clones of each transformation.

Identification of an Activator Element Located Immediately
Upstream of the TATA Box-Our previous studies have shown that a version of the RTBV promoter truncated to position Ϫ50 relative to the transcription start site retained 40% activity of the full-length promoter in transient expression systems (29). To further identify potential upstream cis-acting elements, a series of promoter 5Ј deletion constructs was used to analyze RTBV promoter regulation in transfected rice protoplasts. RTBV promoter sequences to position Ϫ218 were used in these constructs, because the region between positions Ϫ618 (full length) and Ϫ218 does not influence transcription from the RTBV promoter in O. sativa protoplasts (29). All constructs included a full-length RTBV leader sequence and a chloramphenicol acetyltransferase (CAT) ORF fused to RTBV ORF I. As shown in Fig. 1, expression of the CAT reporter gene was not affected by deletion of upstream sequences to position Ϫ70 relative to the transcriptional start site. However, deletion to nucleotide Ϫ40, leaving only the TATA box and transcriptional start site, resulted in a significant loss of CAT activity to a level reflecting only basal transcription. Hence, these results clearly indicated that a cis-regulatory element conferring maximal promoter activity in transfected rice protoplasts is located between the sequences from Ϫ70 to Ϫ40 of the promoter. We refer to this region as activator element (AE).
Multiple Nuclear Proteins Specifically Bind to Sequences from Ϫ60 to Ϫ39 -It has been shown that nuclear proteins from rice seedlings bind to DNA sequences from Ϫ53 to Ϫ39, called box II, of the RTBV promoter (26). To test whether nuclear proteins from rice cell suspension cultures can bind to this activator element, EMSA were performed with nuclear extracts from rice cell suspensions in comparison with rice shoot and root nuclear extracts. One or more DNA-protein complexes were detected upon incubation of each of these extracts with radiolabeled promoter sequences from Ϫ100 to Ϫ1 (Fig. 2). Four specific DNA-protein complexes, designated C1, C2, C3/C3Ј, and C4, were repeatedly observed with nuclear extracts from rice shoots (S) and cell suspensions (C) (Fig. 2, lanes 2 and 3); C3 and C3Ј were observed in shoot nuclear extracts and cell suspension nuclear extracts, respectively. C1 and C2 are predominant complexes with high intensity, and C3Ј migrates slightly faster than C3. However, in root extracts (R), only one DNA-protein complex, named C, was observed with a slightly higher mobility than those in shoot or cell suspension nuclear extracts (Fig. 2, lane 4). Two additional DNA-protein complexes (C5 and C5Ј) that migrate faster than any others were detected with nuclear extract prepared from cell suspension cultures (Fig. 2, lane 3).
To determine the specificity of protein binding to the DNA probe, competition experiments were performed with specific unlabeled DNA fragments. All complexes were completely competed with a 200-fold molar excess of unlabeled Ϫ100 to Ϫ1 probe DNA fragment (fl) (Fig. 3, B-D, lanes 2). A series of shorter double-stranded oligonucleotides covering subregions of this fragment (Fig. 3A) were then used as competitors. The presence of competitor m2 abolished the formation of all complexes (Fig. 3, B-D, lanes 4), while m1 and m3 showed no competition for complexes C1-C4 and C (Fig. 3, B-D, lanes 3  and 5). Complexes C5 and C5Ј can be competed completely by m2 and m3, but only partially by m1 (Fig. 3B, lanes 3, 4, and 5).
These results indicate that nuclear factors bind to the m2 region from Ϫ70 to Ϫ35 to form multiple DNA-protein complexes, and that both m2 and m3 contribute to the formation of complexes C5 and C5Ј. To further determine the minimal sequence requirement for the formation of the major DNA-pro- tein complexes, three overlapping double-stranded oligonucleotides covering different regions from Ϫ70 to Ϫ35 were used as competitors (Fig. 3A). It is evident that m5, which corresponds to the region from Ϫ60 to Ϫ39, is equally effective in competing complexes C1-C4 and C as m2 (Fig. 3, B-D, lanes 7). However, m4 and m6 were not able to compete those complexes (Fig. 3,  B-D, lanes 6 and 8). M6, corresponding to the region from Ϫ70 to Ϫ54, competed for C5 and C5Ј efficiently (Fig. 3B, lane 8), while m5 showed a weak competition (Fig. 3B, lane 7) and m4 none (Fig. 3B, lane 6).
Taken together, these results indicate that the minimal sequence required for formation of DNA-protein complexes C1-C4 and C resides within the region from Ϫ60 to Ϫ39, while the binding sites for C5 and C5Ј complexes are located within the regions from Ϫ70 to Ϫ54 and Ϫ35 to Ϫ1. We did not detect the proteins that bind to the ASL box from Ϫ98 to Ϫ79 reported by Yin et al. (27). The differences in our results may be due to differences in the methods used to prepare nuclear extracts.
Copper-Phenanthroline Footprinting Assays Show That Complexes C1 to C4 Share a Common Binding Site from Ϫ60 to Ϫ39 -Complexes C1 to C4 might bind to individual sites or share the same site within the region from Ϫ60 to Ϫ39. To address this question, we performed copper-phenanthroline footprinting assays. EMSAs were performed with nuclear extracts prepared from cell suspension cultures using the 100-bp probe utilized in Figs. 2 and 3 labeled on either the top or bottom strand. Complexes corresponding to C1, C2, C3Ј, C4, C5, and C5Ј were then subjected to footprinting analysis. DNA from each fraction was recovered and characterized on sequencing gels. The footprinting patterns for the top and bottom strands of the RTBV promoter are shown in Fig. 4A. A single protected area was detected from Ϫ60 to Ϫ39 in both the top and bottom strands from complexes C1, C2, C3Ј, and C4 (Fig.  4C). The large footprints observed probably result from the binding of more than one protein. A single protected area in the top strand and two protected areas in the bottom strand were detected from complex C5 (mixture of C5 and C5Ј) (Fig. 4A). The protected areas were located in a region from Ϫ70 to Ϫ60 in the top strand and in regions from Ϫ60 to Ϫ54 and Ϫ50 to Ϫ43 in the bottom strand (Fig. 4C). The footprinting data are in precise agreement with the results of the binding assays performed with competitors and indicate that all complexes except C5 and C5Ј share a common binding site including nucleotides from Ϫ60 to Ϫ39. In all complexes, the A at position Ϫ96 showed an enhanced reactivity on the top strand in the presence of protein, while unbound DNA (F) is almost not cleaved at this residue under the conditions employed (Fig. 4A).
Methylation Interference Analysis Shows That Proteins Contact the Activator Element in the Major Groove-To further identify the precise contacts made by nuclear proteins in binding to this region of DNA, methylation interference analysis was performed using the Ϫ100 to Ϫ1 probe labeled on either the top or bottom strand. Partially methylated DNA bound to nuclear proteins was cleaved at G residues methylated at the N-7 position, which interferes with major groove DNA binding (39). Methylation of guanine nucleotides in the top strand at positions Ϫ55 and Ϫ46 and in the bottom strand at positions Ϫ53, Ϫ45 to Ϫ42 strongly affected the specific DNA-protein interactions (Fig. 4B). Methylation of the guanine nucleotide in the bottom strand at position Ϫ52 resulted in a weak interference with the protein binding. Again, complexes C1, C2, C3Ј, and C4 all involved the same G residue contacts in both top and bottom strands (Fig. 4C). The G at position Ϫ55 in the top strand was not involved in protein binding analyzed previously by DNase I footprinting (26). To verify the involvement of the G at position Ϫ55 in protein binding, a G to T mutation at this position (within the context of m2, Ϫ70 to Ϫ35, see Fig. 3A) was tested as a competitor in EMSA. The mutated sequence was not able to compete any of the complexes (data not shown), suggesting that the G at position Ϫ55 is crucial for formation of the DNA-protein complexes. Taken together, the results indicated that nuclear proteins directly contacted the activator element in the major groove.

The Formation of Multiple DNA-Protein Complexes on the Activator Element Is Based on Protein-Protein Interactions-
Because multiple DNA-protein complexes are observed in EMSA and only a single protected area is detected in footprinting analysis from complexes C1, C2, C3Ј, and C4, we hypothesized that these multiple complexes result from either proteinprotein association of a common DNA-binding protein with non-DNA-binding proteins or the interaction of multiple DNAbinding proteins of different sizes to a common binding site. To test this hypothesis, we utilized the detergent DOC in EMSA, which is known to disrupt protein-protein interactions and has been widely used to characterize multi-protein complexes (40 -43). When DOC was added in low concentration, the four DNAprotein complexes were converted into only two complexes that migrated faster than the initial complexes with nuclear extracts from cell suspension cultures and shoots (Fig. 5). Adding an excess of the nonionic detergent Nonidet P-40 reversed this effect (Fig. 5, lanes 5 and 10). This observation implies that protein-protein interactions were required for the formation of multiple DNA-protein complexes, and that there are two specific DNA-binding proteins, designated AEBP-1 and AEBP-2 (activator element-binding protein 1 and 2), respectively, directly recognizing the AE. The DOC treatment abolished the formation of complexes C5 and C5Ј even at very low concentration (Fig. 5A, lane 2), and complex formation could not be restored by Nonidet P-40 (Fig. 5A, lane 5), suggesting that proteins interact with each other and are unable to bind DNA alone.
Identification of Proteins Bound to the AE by DNA UV Cross- linking Assays-Our results demonstrate that formation of multiple DNA-protein complexes on the AE is based on proteinprotein interactions, and only two proteins directly contact this element. To specifically identify these proteins, we performed DNA UV cross-linking assays with a bromo-dUTP-labeled probe. The bromo-dUTP-labeled probe formed the same specific DNA-protein complexes (data not shown) as the regular probe Ϫ100 to Ϫ1 (Fig. 2). After UV cross-linking, DNA-protein complexes were analyzed by SDS-polyacrylamide gel electrophoresis. In addition to a nonspecific protein of 30 kDa (see below), which appeared in all extracts tested, two proteins with apparent molecular masses of 33 and 36 kDa were identified in nuclear extracts from cell suspension cultures and shoots, while only one protein of 26 kDa was detected in root nuclear extracts (Fig. 6). The 36-kDa protein predominates in shoot nuclear extracts, while the 33-kDa protein is more abundant than the 36-kDa protein in cell suspension culture nuclear extracts.
Competition experiments were performed in the presence of a 200-fold molar excess of three different unlabeled DNA com-petitors (m1, m2, and m3; see Fig. 3A). The 30-kDa band was not competed by any of these competitors; therefore, it was judged to be nonspecific. All the other cross-linked proteins were competed by competitor m2 (corresponding to region Ϫ70 to Ϫ35), but not by m1 and m3, demonstrating specific interactions between the AE and these cross-linked proteins (Fig. 6). These results are consistent with those determined by EMSA. Treatment of the UV-irradiated reaction mixture with proteinase K prior to gel electrophoresis abolished complex formation, indicating that nuclear proteins were indeed present in the bands detected (data not shown). We could not detect proteins that form complexes C5 and C5Ј with the nuclear extract prepared from cell suspension culture. Possibly, these might be present in the extract at levels below the limit of detection using this method.
The AE Is a Prerequisite for Promoter Activity-To address the functional consequence of the DNA-protein interactions with the AE, deletions or mutations were made in the context of the whole leader and promoter sequences to Ϫ218. The activity of these constructs was assessed in transfected rice protoplasts. Internal deletion from Ϫ70 to Ϫ35 resulted in a dramatic reduction in promoter activity to less than 10% of wild type (Fig. 7). Mutation of Gs identified as being directly contacted by nuclear factors to Ts had a similar effect. A double-stranded oligonucleotide from Ϫ70 to Ϫ35 bearing these mutations was also not able to compete any of the DNA-protein complexes when used as a competitor in EMSA (data not shown). These results highlight the crucial importance of the AE for efficient transcription from the promoter.
The Position of the AE Relative to the TATA Box Is Critical for Promoter Activity-To determine whether the AE could confer promoter activity from an upstream position, sequences from Ϫ70 to Ϫ35 were deleted in the wild-type position and one copy of this DNA fragment was placed at position Ϫ218 upstream of the transcription start site. As shown in Fig. 7, one copy of the AE positioned at nucleotide Ϫ218 in either forward or reverse orientation is not able to compensate at all for the deletion of sequences from Ϫ70 to Ϫ35. This finding indicated that the AE could not enhance promoter activity from a distance. Thus, activation by the AE is position-dependent.
Functional Association of the AE Is DNA Distance-dependent but Not Turn-dependent-Inspection of the sequences in the promoter-proximal region revealed that the binding sites for nuclear factors within the AE are very close to the TATA box, with only a very few base pairs separating the two elements. This close proximity, and the position-dependent characteristics of the AE raised the question whether AE-binding activator proteins would interact directly or indirectly with basal transcription factors to activate transcription from the promoter. If so, it would be predicted that altering the spacing between the AE and the TATA box could affect protein-protein interactions between these two sets of factors, and thereby influence transcription from the promoter. To test this prediction, a series of constructs was made in which the distance between the AE and the TATA box was varied from 5 to 20 bases (Fig. 8). The effect on promoter activity was tested in transfected rice protoplasts. As the distance between the AE and the TATA box increased, the promoter activity gradually decreased (Fig. 8). When five nucleotides, corresponding to half a DNA helical turn, were inserted between the AE and the TATA box, CAT activity was reduced to 73% of the wild-type construct, indicating a lack of strict dependence on stereospecific alignment between the AE and the TATA box for activation. A further reduction was observed with a 10-nucleotide insertion (a full DNA helical turn). Insertion of 15 or 20 nucleotides resulted in a significant loss of promoter activity (to 32% and 26% of wild-type, respectively). Thus, activation via the AE is DNA distance-dependent but not turn-dependent.
Loss of Promoter Activity Is Not Due to the Loss of Protein Binding to the AE-To determine whether these insertions affect the formation of DNA-protein complexes on the AE, EMSA was performed using DNA probe bearing the same insertions between Ϫ35 and Ϫ34 as in the above constructs. The DNA-protein complexes with these probes in nuclear ex- FIG. 4. Precise localization of the nuclear factor binding sites within the AE on the promoter. A DNA fragment spanning Ϫ100 to Ϫ1 was 5Ј end-labeled on either the top or bottom strand and incubated with cell suspension nuclear extracts. A, copper-phenanthroline footprinting analysis. Labeled DNA probe was incubated with nuclear extracts and resolved on a native polyacrylamide gel. DNA-protein complexes (C1-C5) as well as free probe (F) were digested in situ with 1,10-phenanthroline-copper ion, eluted from the gel, and complexes resolved on a 6% polyacrylamide sequencing gel. A Maxam and Gilbert sequencing ladder (G) was run in parallel. The protected areas are depicted on the right of the gel by open (C1-C4) or filled (C5) vertical bars. The numbers on the right correspond to base pairs upstream of the transcription start site. A strong hypersensitive site at nucleotide A at position Ϫ96 is indicated with an arrow. B, methylation interference footprinting reveals that protein contacts are made in the major groove of the AE. A labeled DNA probe was partially methylated with dimethyl sulfate, incubated with nuclear extracts, and complexes resolved on a native polyacrylamide gel. Both protein-bound (complexes C1-C4) and free (F) DNA were eluted from the gel, cleaved at the modified bases with piperidine, and resolved on a 6% polyacrylamide sequencing gel. A Maxam and Gilbert sequencing ladder (G) was run in parallel. The positions of G residues at which methylation interfered with binding are indicated with arrowheads. tracts from both cell suspension cultures and shoots were identical to those obtained with wild-type probe, and the resulting DNA-protein complexes can be competed completely by wildtype sequences from Ϫ70 to Ϫ35 (data not shown). These results indicated that these insertions did not affect DNA-protein interactions on the AE, and therefore loss of promoter activity is not due to loss of protein binding to the AE. DISCUSSION To elucidate the molecular mechanisms involved in regulating transcription from the RTBV promoter, we have undertaken a systematic analysis of the AE and its cognate transacting factors. The current study provides a new insight into the transcriptional regulation mechanisms of RTBV. We have demonstrated that (i) nuclear proteins bind to the AE, forming multiple DNA-protein complexes through protein-protein interactions, (ii) AE is a position-dependent element whose function is critically dependent on its position relative to the TATA box, (iii) AE is a prerequisite for efficient promoter activity, and (iv) activators probably stimulate transcription via interactions with basal transcription factors.
Formation of multiple DNA-protein complexes on the AE is based on protein-protein interactions, and these interactions correlate with the functional activity of the promoter. EMSA analyses with different nuclear extracts have led to the identification of different DNA-binding proteins, which may be involved in regulating cell type-specific transcription from the promoter. Our EMSA results showed that the protein binding patterns are not exactly the same in cell suspension, shoot and root nuclear extracts, indicating that different proteins from different types of nuclear extracts directly or indirectly bind to the same element. DNA UV cross-linking studies revealed two proteins of 36 and 33 kDa binding to this element, suggesting that formation of protein-protein interactions in the RTBV promoter requires the binding of both the 36-and 33-kDa proteins to this element. Based on the protein profile and the mobility of each complex in native gels, it is most likely that AEBP-1 is the 33-kDa protein, while AEBP-2 is the 36-kDa protein. Complex C1 contains two DNA-binding proteins of 33 and 36 kDa, and complexes C2, C3, C3Ј, and C4 may represent heterodimers of the 33-or 36-kDa proteins with unknown proteins, which do not directly bind to DNA. However, a 26-kDa DNA-binding protein was found in root nuclear extracts with which no protein-protein interaction was observed. Taken together, we may conclude that the direct or indirect binding of different nuclear factors to the same element may contribute to alternative transcriptional regulation mechanisms from the promoter in different cell types, and that protein-protein interactions play a critical role in transcriptional activation from the promoter.
The importance of the AE for transcription from the RTBV promoter in vivo is apparent from the effects of internal deletions and mutations in this element on the activity of the promoter tested in a transient expression system (Fig. 7). Moreover, when the DNA fragment from Ϫ70 to Ϫ35 was placed upstream of promoter at position Ϫ218, it was not able to exert its effect regardless of orientation, suggesting that its function is position-dependent. Mutation of Gs directly contacted by nuclear proteins in both top and bottom strands resulted in the same reduction of promoter activity as deletion of this element. These mutations also disrupt the formation of DNA-protein complexes. Taken together, these results indicated that the AE is a position-dependent element, and that there is a complete correlation between nuclear protein binding in vitro and transcription activation in vivo. Interestingly, de-letion or mutation of this element has a more drastic effect than deletion of upstream sequences to Ϫ40 (cf. Figs. 1 and 7). This is most likely due to an inhibitor element located upstream at sequences from Ϫ165 to Ϫ100 identified in transfected rice protoplasts. 2 Studies on the RTBV promoter in transgenic rice plants have shown that three upstream elements, box II (Ϫ53 to Ϫ39), the ASL box (Ϫ98 to Ϫ79), and the GATA motif (Ϫ143 to Ϫ135), act combinatorially to confer phloem-specific gene expression in the context of the Ϫ164 to ϩ45 promoter (27), and that a promoter element located to sequences from Ϫ164 to Ϫ100 is very important for gene expression in the vascular tissue (28). The results presented here show that, for full promoter activity, upstream sequences to Ϫ70 containing the activator element in the context of complete leader are required in transfected rice protoplasts. The differences in our results may reflect differences between stable and transient expression systems (28,30), as demonstrated by EMSA analysis in this study.
In the ␣-amylase gene promoter regulated by gibberellin (GA) during germination, a cis-acting element called box T is located three bases upstream of the TATA box. Box T is a 2 X. He, J. Fü tterer, and T. Hohn, unpublished results. , and Ϫ45 to Ϫ42) strands to Ts. All constructs were tested in at least three independent transfections. For each construct, the mean promoter activity is indicated as a percentage of the activity of the wild-type construct (set to 100%).
FIG. 8. Altering the distance between the AE and the TATA box significantly reduces promoter activity. A schematic representation of the constructs tested in transfected O. sativa protoplasts is shown on the top left, with the expanded sequence underneath indicating insertion of synthetic oligonucleotides carrying one, two, three, or four copies of AATTG between positions Ϫ35 and Ϫ34. The TATA box is underlined. All constructs were tested in at three independent experiments. For each construct, the mean promoter activity is indicated as percentage of the activity of the wild-type construct (set to 100%). polypyrimidine sequence element ( Ϫ50 GATCACATC-CCCCCT Ϫ36 ). It was suggested that box T may be bound by basal transcription factors (44). The FP56 element in the parsley 4CL1 promoter is also a pyrimidine-rich element (5Ј-TC-CCCATTTACCCCT-3Ј) on which multiple DNA-protein complexes form (45). This element consists of a perfect indirect repeat of the octanucleotide TCCCCATT, but its functional significance remains to be clarified. The AE is also pyrimidinerich and is apparently not bound directly by basal transcription factors. Inspection of the AE nucleotide sequence revealed the presence of two motifs: a TGACG-like motif ( Ϫ56 TGACC Ϫ51 ) and an octamer-like motif ( Ϫ48 GTGCCCCT Ϫ41 ). This octamer sequence is found in the C4-type phosphoenolpyruvate carboxylase gene in maize and binds a light-inducible DNA-binding factor (MUF1) (46). The TGACG motif and variations have been identified in the promoters of a variety of plant genes, such as the auxin-regulated tobacco glutathione transferase genes Nt103-1 and Nt103-35 (47), the CaMV 35 S promoter (the activation sequence-1 element) (48,49), the octopine synthase gene (50), a wheat histone gene (14), a light-regulated TGA1 gene (51), a cucumber hpr-A gene (52), and a tobacco group 1 pathogenesis-related proteins (PR-1) gene (53).
Recently, a gene encoding the rice bZIP transcription factor RF2a that is critical for vascular development was isolated (54). This factor was found to interact with RTBV promoter elements located between nucleotides Ϫ53 and Ϫ39. The factor was not present in the DNA-protein complexes observed in this study.
A recent report analyzing transcriptional activation by the Arabidopsis transcription factor GT-1 showed that the factor can interact with and stabilize the TFIIA-TBP-TATA complex, suggesting that GT-1 may activate transcription through direct interaction with the transcriptional pre-initiation complex (18). Activators often lose activity as their distance from the initiation site is increased (55), since activators must be located stereospecifically with respect to the TATA box for transcriptional activation (56 -58). However, in some instances, the alignment of the DNA binding site was shown not to play a strict role in promoter activity (59). The strict requirement for stereospecific alignments of protein binding sites in the promoter regions does not seem to hold in the RTBV case. However, the location of the AE relative to the TATA box is very important for transcription activation, indicating that the relative spacing rather than DNA helical turn is most important for activation. The simplest interpretation of these data is that there is lateral flexibility of the binding proteins. A model for transcriptional activation by activators in the RTBV promoter is presented in Fig. 9. This model involves two activators (P36 and P33) binding directly to the AE, unknown proteins then interact with P36 and P33 through protein-protein interactions. For transcriptional activation, activators recruit holoenzyme containing TFIID to the core promoter. On the promoter, the activators and TFIID are brought into proximity, enabling proteins that directly or indirectly bind to the activator element to interact with TFIID. The interactions between activators and TFIID can facilitate and stabilize the transcriptional preinitiation complex, and thereby stimulate transcription. As the distance between the bound activators and bound TFIID is increased, this interactive effect is diminished, leading to a decrease in transcription. In the roots, no protein-protein interactions were observed. The low transcription level detected in roots of transgenic rice plants (26 -28) may be due to the fact that the activator is not able to interact with basal transcription factors in this tissue. Isolation of cDNAs expressing these DNA-binding proteins, either by screening an expression library using the minimal binding sequence or by protein purification, will help us to verify the above model and further elucidate transcriptional regulation mechanisms from RTBV promoter.