Detailed Architecture of the Barley ChloroplastpsbD-psbC Blue Light-responsive Promoter*

The photosystem II reaction center chlorophyll protein D2, is encoded by the chloroplast gene psbD. PsbDis transcribed from at least three different promoters, one which is activated by high fluence blue light. Sequences within 130 base pairs (bp) of the psbD blue light-responsive promoter (BLRP) are highly conserved in higher plants. In this study, the structure of thepsbD BLRP was analyzed in detail using deletion and site-directed mutagenesis and in vitro transcription. Deletion analysis showed that a 53-bp DNA region of thepsbD BLRP, from −57 to −5, was sufficient for transcription in vitro. Mutation of a putative prokaryotic −10 element (TATTCT) located from −7 to −12 inhibited transcription from the psbD BLRP. In contrast, mutation of a putative prokaryotic −35 element, had no influence on transcription. Mutation of a TATATA sequence located between the barley psbA −10 and −35 elements significantly reduced transcription from this promoter. However, site-directed mutation of sequences located between −35 and −10 had no effect on transcription from the psbDBLRP. Transcription from the psbD BLRP was previously shown to require a 22-bp sequence, termed the AAG-box, located between −36 and −57. The AAG-box specifically binds the protein complex AGF. Site-directed mutagenesis identified two different sequence motifs in the AAG-box that are important for transcription in vitro. Based on these results, we propose that positive factors bind to the AAG-box and interact with the chloroplast-encoded RNA polymerase to promote transcription from the psbD BLRP. Transcription from the psbD BLRP is thus similar to type II bacterial promoters that use activating proteins to stimulate transcription. Transcription of the psbD BLRP was ∼6.5-fold greater in plastid extracts from illuminated versus dark-grown plants. This suggests that light-induced activation of this promoter in vivo involves factors interacting with the 53-bp psbDBLRP in vitro.

Photosystem II contains at least four plastid-encoded chlorophyll apoproteins (D1, D2, CP47, CP43). Among these, D2 and D1 form a heterodimer, which houses the photosystem II reaction center chlorophyll P680. D1 and D2 are relatively unstable in illuminated plants (1)(2)(3)(4)(5). Therefore, synthesis of D1 and D2 is selectively elevated in mature barley chloroplasts in order to maintain the levels of these subunits and PSII function (5,6). Maintenance of high rates of D1 and D2 synthesis in mature barley chloroplasts is paralleled by the retention of elevated levels of psbA and psbD mRNAs, which encode these proteins (6 -8). D1 mRNA levels remain high in mature barley chloroplasts primarily due to the high stability of its mRNA, although transcription from psbA is also increased by light (9 -12). Maintenance of high levels of psbD mRNA results primarily from the activation of psbD transcription by blue light combined with a small increase in RNA stability (5,13).
The chloroplast genome in most higher plants is circular and ranges in size from 120 to 217 kilobase pairs (reviewed in Refs. 14 -17). The genome encodes approximately 135 genes including genes for rRNAs, tRNAs, subunits of the plastid 70 S ribosome, subunits of an RNA polymerase (rpoA, rpoB, rpoC1, and rpoC2), and proteins that comprise the photosynthetic apparatus. Transcription of the chloroplast genome is complex and highly regulated (reviewed in Refs. 17 and 18). Plastid genes are transcribed by at least two different RNA polymerases (RNAPs). 1 The catalytic subunits of one RNAP are encoded by the chloroplast genes rpoA, rpoB, and rpoC1/C2 (reviewed in Ref. 19). This RNAP recognizes prokaryotic Ϫ10 and Ϫ35 promoter elements (reviewed in Ref. 18). Other types of plastid promoters have been identified. For example, the promoter for the rps16 gene contains only a Ϫ35 element (20). Other genes, such as trnS, trnR (21), rpoB (22), rpl32 (23), and rpl23 (24) are not preceded by typical prokaryotic promoter consensus elements. Many of these genes are transcribed by a nucleus-encoded RNAP (Refs. 22,25,and 26;reviewed in Ref. 17). This polymerase is likely to be encoded by the nuclear gene rpoZ, which shows sequence similarity to the bacteriophage T7 and SP6 RNA polymerases (27). Plastid transcription is also regulated via multiple -factors (28 -30), which may be phosphorylated (31,32). Other DNA binding complexes, such as CDF2 and AGF, have been identified, which modulate transcription of rrn (33), and psbD-psbC (34), respectively.
In barley, psbD is located in a complex operon that also contains psbC, psbK, psbI, orf62, and trnG (35). The psbD operon is transcribed from at least three different promoters (13). One of the psbD promoters is activated when plants are illuminated by high fluence blue light but not by red or far-red illumination (5,36). Transcripts arising from the blue lightresponsive promoter (BLRP) become the most abundant psbD transcripts in chloroplasts of mature barley leaves (13,37). Light-induced accumulation of psbD transcripts has been observed in a wide variety of plants (37)(38)(39). A ϳ130-bp region surrounding the psbD BLRP is conserved among cereals, dicots, and black pine (34,37) despite DNA rearrangements upstream of the psbD BLRP in some plants (37). The conserved psbD BLRP contains sequences with significant similarity to typical prokaryotic Ϫ10 and Ϫ35 promoter regions (13). In addition, two conserved regions, termed the AAG-box and PGTbox, are located upstream of the putative Ϫ35 element (34). Previously, we showed that the AAG-box and its cognate DNAbinding protein complex, AGF, are required for transcription from the barley psbD BLRP in vitro (34). Furthermore, the DNA region containing the PGT and AAG-boxes was shown to be important for transcription from the tobacco psbD BLRP in vivo (40). In the present study, we define a minimal DNA region required for transcription of the barley psbD BLRP and further dissect the architecture of the promoter using deletion, insertion, and point mutation analyses.

EXPERIMENTAL PROCEDURES
Plant Growth-Barley (Hordeum vulgare var. Morex) seedlings were grown in controlled environmental chambers at 23°C as described by Kim et al. (12). Seedlings were germinated and grown in complete darkness. After 7.5 days, the dark-grown seedlings were either harvested or transferred to a continuously illuminated chamber (fluorescent plus incandescent light, light intensity 250 microeinsteins m Ϫ2 s Ϫ1 ) for an additional 16 h before harvesting. Plastids were isolated from the top 5-7 cm of primary leaves of barley seedlings by Percoll gradient (35-75%) centrifugation (41). The concentration of plastids was quantitated (plastids per microliter) by phase contrast microscopy using a hemacytometer.
Preparation of Plastid Extracts for in Vitro Transcription Experiments-The plastid high salt extracts used for in vitro transcription experiments in this study were prepared according to Kim and Mullet (34). Protein extract from 5.2 ϫ 10 8 plastids obtained from approximately 7.5-day-grown barley plants was used in each in vitro transcription assay.
In Vitro Transcription and Primer Extension Analyses-Transcription of exogenous DNA templates in vitro and primer extension analyses of in vitro transcribed DNA were performed as described by Kim and Mullet (34). The minus 40 primer was used to analyze transcripts originating from the plasmid pLRP140 and its derivative recombinant plasmids; the T3 primer was used for ppsbA138, prbcL216, and their derivative recombinant plasmids.
Plasmid Construction of pLRP97, pLRP80, pLRP69, and pLRP121-Recombinant plasmids pLRP97, pLRP80, pLRP69, and pLRP121 were constructed based on PCR cloning, using pLRP185 (34) as a DNA template. The 3Ј-end primers used for amplification of LRP97, LRP80, and LRP69 were designed based on cDNA-like sequences as follows: LRP97, 5Ј-TTCGCggATcCAATTTCATCTAC (ϩ33 to ϩ11 region of the psbD BLRP); LRP80, 5Ј-TTCATCTggATCcAATTTATATA (ϩ19 to Ϫ4 region of the psbD BLRP); LRP69, 5Ј-GAATTggatccTCAGAAT-AGCGGA (ϩ7 to ϩ17 region of the psbD BLRP). Engineered BamHI sites are underscored, and nucleotide changes from native chloroplast sequences are designated by lowercase letters. The 5Ј-end primer used for PCR amplification of the three DNA fragments mentioned above was based on mRNA-like sequences as follows: 5Ј-TCAAATCtAgaATA-AAATTGGAAA (Ϫ81 to Ϫ62 region of the psbD BLRP, engineered XbaI site underscored with nucleotide changes from native chloroplast sequences designated by lowercase letters).
The 5Ј-and 3Ј-end primers used for PCR amplification of LRP121 are as follows: 5Ј-end primer, 5Ј-ATTGGtctAgaCATAAAGTAAGTA (mRNA-like sequences, Ϫ68 to Ϫ45 region of the psbD BLRP); 3Ј-end primer, 5Ј-TTCGCggATcCAATTTCATCTAC (cDNA-like sequences, ϩ33 to ϩ11 region of the psbD BLRP). Both of the engineered BamHI and XbaI sites are underscored, and nucleotide changes from native chloroplast sequences are shown with lowercase letters. All of the PCR products mentioned above were digested by BamHI and XbaI, gelpurified, and ligated into BamHI and XbaI sites of pBluescript SKϩ. The resulting plasmids were named pLRP97, pLRP80, pLRP69, and pLRP121, respectively.
Site-directed Mutagenesis of Ϫ10, Ϫ35, and TATA Elements in pLRP140, prbcL216, and ppsbA138 -Site-specific base substitution in Ϫ10, Ϫ35, and TATA elements in pLRP140, prbcL216, and ppsbA138 (34) was introduced based on a PCR-based "overlap extension technique" described by Higuchi et al. (42) and Ho et al. (43). Previously, prbcL216 was constructed by inserting a PCR-amplified DNA fragment, extending from Ϫ156 to ϩ60, flanking the transcription initiation site of the barley rbcL (44), into BamHI and EcoRI sites of pBluescript SKϩ. The inside primers, coupled with the outside primers, to generate two overlapping primary PCR fragments, which bear the same mutations in the region of overlap, are as follows: pLRP140/Ϫ35mt, 5Ј-TGACTCCaT-cAATGATGCCT for upper strand primer, 3Ј-ACTGAGGtAgTTAC-TACGGA for lower strand primer (Ϫ40 to Ϫ21 region of the psbD BLRP, respectively); pLRP140/Ϫ10mt, 5Ј-TATCCGCaATTCaGATATAT for upper strand primer, 3Ј-ATAGGCGtTAAGtCTATATA for lower strand primer (Ϫ19 to ϩ1 region of the psbD BLRP, respectively); pLRP140/ Ϫ35&Ϫ10mt, 5Ј-TCCaTcAATGATGCCTCTATCCGCaATTCaGAT for upper strand primer, 3Ј-AGGtAgTTACTACGGAGAATAGGCGtTA-AGtCTA for lower strand primer (Ϫ36 to Ϫ4 region of the psbD BLRP, respectively); prbcL216/Ϫ35mt, 5Ј-ATTTGGGaTcCGCTATACCT for upper strand primer, 3Ј-TAAACCCtAgGCGATATGGA for lower strand primer (Ϫ41 to Ϫ22 region of the barley rbcL, respectively); prbcL216/ Ϫ10mt, 5Ј-CAAGAGTAaACAAaAATGATGG for upper strand primer, 3Ј-GTTCTCATtTGTTtTTACTACC for lower strand primer (Ϫ19 to ϩ4 region of the barley rbcL, respectively); ppsbA138/Ϫ35mt, 5Ј-TGACTT-GGaTcACATTGGTATA for upper strand, 3Ј-ACTGAACCtAgTGTAAC-CATAT for lower strand primer (Ϫ45 to Ϫ19 region of the barley psbA, respectively); ppsbA138/Ϫ10mt, 5Ј-GTCTATGTaATACaGTTAAATA for upper strand primer, 3Ј-CAGATACAtTATGtCAATTTAT for lower strand primer (Ϫ21 to ϩ1 region of the barley psbA, respectively (45)); ppsbA138/TATAmt, 5Ј-GACATTGGaAgAaAGTCTATGT for upper strand primer, 3Ј-CTGTAACCtTcTtTCAGATACA for lower strand primer (Ϫ35 to Ϫ14 region of the barley psbA, respectively); ppsbA138/ Ϫ35&TATAmt, 5Ј-TGACTTGGaTcACATTGGaAgAaAGTCTAT for upper strand primer, 3Ј-ACTGAACCtAgTGTAACCtTcTtTCAGATA for lower strand primer (Ϫ45 to Ϫ16 region of the barley psbA, respectively). The specific base substitutions introduced in each primer, which create mismatch between a primer and the individual template target sequence, are designated by lowercase letters. T3 and T7 primers, which hybridize to the regions originating from the vector sequences of pBluescript SKϩ in pLRP140, prbcL216, and ppsbA138, respectively, were used together with the primers described above to generate the overlapping primary PCR fragments. Each set of overlapping primary PCR products was gel-separated, mixed together, denatured, and allowed to reanneal. Each resulting extended segment was then used for the secondary amplification of the combined sequences, using the outside T3 and T7 primers, which were employed to produce the primary fragments. After the secondary PCR amplification, XbaI and XhoI restriction enzyme sites were used to insert the individual DNA fragments into pBluescript SKϩ.

RESULTS
Minimal DNA Region Required for Transcription from the psbD BLRP-The structure of the barley chloroplast psbD BLRP is shown in Fig. 1A. Comparisons of the psbD BLRP region among numerous plants showed several stretches of sequence conservation from approximately ϩ30 to Ϫ100. In particular, sequences surrounding the AAG-box (Ϫ36 to Ϫ64) and PGT-box (Ϫ71 to Ϫ100) are highly conserved (34,37). Previous analysis of the psbD BLRP demonstrated that the region from ϩ64 to Ϫ76 was sufficient to activate transcription in vitro (34). In this study, sequences important for transcription in vitro were further delineated using a series of deletions of pLRP140 (Fig. 1A).
Recombinant plasmids pLRP97, pLRP80, and pLRP69 contain a series of 3Ј-end deletions of the psbD BLRP (Fig. 1A). Each of the recombinant plasmids was added to chloroplast in vitro transcription extracts obtained from 8-day-old barley plants that had been illuminated for 16 h. The psbD transcripts produced from the plasmids were assayed using primer extension analysis. No psbD transcript 5Ј termini were observed when mock transcription reactions were analyzed (data not shown). However, as shown in Fig. 1B (lanes 1-4), all of the 3Ј deletion recombinant plasmids and pLRP140 were equally good templates. Similar results were observed with plastid extracts from 7.5-day-old, dark-grown barley plants (data not shown).
Previous analyses demonstrated that the sequence AAAG-TAAG (Ϫ54 to Ϫ47) in the AAG-box (see Fig. 1A) was required for transcription from the BLRP (34). To examine the influence of sequences upstream of this sequence, pLRP121 was constructed, which contains a 5Ј deletion to Ϫ57 (Fig. 1A). This deletion caused no loss of transcription activity from the psbD The boxed regions identify conserved sequences including the AAGbox, the PGT-box, and sequences homologous to E. coli Ϫ35 and Ϫ10 promoter elements (34). The site of psbD transcription initiation is designated by an arrow, and labeled as ϩ1. Repeated sequences in the AAG-box are underlined. pLRP140 is a recombinant plasmid containing 140 bp of DNA (Ϫ76 to ϩ64) from the psbD BLRP (34). pLRP185 was used as a template to construct recombinant plasmids, pLRP97, pLRP80, pLRP69, and pLRP 121. B, in vitro transcription of the recombinant plasmids shown in A using barley plastid extracts. Transcripts were analyzed using primer extension analysis. The arrows designate primary transcripts produced from the recombinant plasmids. The asterisk marks the position of a signal produced from plastid extracts in the absence of template. DNA size markers in base pairs are indicated to the right. BLRP (Fig. 1B, lane 5). These results indicate that the ϳ53-bp DNA region from Ϫ57 to Ϫ5 is sufficient for transcription from the psbD BLRP in vitro.
The sites of transcription initiation from the 3Ј deletion constructs of pLRP140 were fine mapped using primer extension analysis (Fig. 2). The 5Ј termini of the psbD transcripts produced from pLRP97 and pLRP80 mapped seven nucleotides downstream from a potential Ϫ10 promoter element (TATTCT) at the same site as the 5Ј terminus of the transcript produced from pLRP140 (34). Similarly, transcripts produced from pLRP69, which contained a 3Ј deletion to Ϫ5, mapped seven nucleotides downstream from the TATTCT Ϫ10 sequence although native nucleotides from ϩ1 to Ϫ4 (TATAT) had been deleted and replaced by CTAGG. This indicates that the sequences immediately surrounding the site of transcription initiation can be modified with minimal influence on transcription initiation.
Analysis of Putative Prokaryotic Ϫ35 and Ϫ10 Promoter Sequences-The psbD BLRP contains potential prokaryotic Ϫ35 (TTGAAT) and Ϫ10 (TATTCT) promoter elements, located at positions Ϫ28 to Ϫ33 and Ϫ7 to Ϫ12, respectively (see Fig. 1A). These sequences are separated by 15 bp. Similar prokaryotic promoter elements, which are separated by 18 bp, have previously been identified upstream of the sites of transcription initiation in the rbcL and psbA promoters (reviewed in Ref. 16). In addition, a TATATA sequence located between the psbA Ϫ10 and Ϫ35 promoter elements contributes to promoter activity in mustard (46).
The function of the putative Ϫ10 and Ϫ35 prokaryotic promoter elements present in the psbD BLRP was analyzed by site-directed mutagenesis. As a control, the influence of modifying the Ϫ10 and Ϫ35 elements in the rbcL and psbA promoters was examined to ensure the in vitro transcription extract was faithfully replicating previous results. Our general approach was to introduce point mutations in potential Ϫ35 and Ϫ10 sequences at sites that show the highest conservation in both plastid and bacterial promoters (Ϫ35, TTGaca; Ϫ10, TAtaaT) (Ref. 47; reviewed in Refs. 16 and 48 -51). The first and the third nucleotides (T and G) in the potential Ϫ35 promoter element of each promoter were switched to A and C, respectively (Figs. 3A and 4A). In the case of potential Ϫ10 promoter elements, the first and the sixth nucleotides, T and T, were both switched to A (Figs. 3A and 4A). The point mutations described above did not create any other potential Ϫ35 or Ϫ10 promoter elements. Each of the recombinant plasmids containing the point mutations was added to plastid transcription extracts, which were obtained from either 7.5-day-old, darkgrown barley plants, or similar plants that had been further illuminated for 16 h.
Transcription from the psbA and rbcL promoter constructs is shown in Fig. 3, B and C. Transcription from the wild type psbA or rbcL promoters was active in extracts of etioplasts isolated from dark-grown plants or chloroplasts isolated from illuminated plants (Fig. 3, B and C, lanes 1 and 2). Modification of the Ϫ35 sequences in these two promoters caused transcription to decrease to very low levels (Fig. 3, B and C, lane 3). Similarly, modification of the Ϫ10 sequences also caused transcription to decrease significantly (Fig. 3, B and C, lane 4). When the TATATA sequence located between the Ϫ35 and Ϫ10 elements in psbA was modified, transcription was reduced although not eliminated (Fig. 3B, lane 5). Not surprisingly, modification of both the Ϫ35 and TATATA sequence in the psbA promoter reduced transcription to nondetectable levels (Fig. 3B, lane 6).
The results in Fig. 4 show the influence of mutation of putative Ϫ10 and Ϫ35 promoter elements found in the psbD BLRP. Mutation of the Ϫ35 sequences did not alter transcription from the psbD BLRP (Fig. 4B, lane 2 versus lane 3). In contrast, mutation of the prokaryotic Ϫ10 element, TATTCT, reduced transcription from the psbD BLRP to very low levels (Fig. 4B, lane 2 versus lane 4). As expected, point mutations in both of the Ϫ10 and Ϫ35 sequences also abolished transcription (Fig. 4B, lane 5). These results indicate that the Ϫ10 sequence is required for transcription from the psbD BLRP in vitro, whereas the Ϫ35 and Ϫ10 elements are both required for transcription from the rbcL and psbA promoters. In addition, the psbA TATATA sequence is important for transcription from the psbA promoter in barley. Transcription of all constructs was greater in extracts of plastids from illuminated plants compared with extracts from etioplasts of dark-grown plants, although the influence of illumination was greatest on the psbD BLRP (ϳ6.5-fold versus 4-fold (psbA) and 2-fold (rbcL) (Figs. 3 and 4)).
Role of the Sequence and Spacing between the AAG-box and Ϫ10 Sequence in the psbD BLRP-In the psbA promoter, a TATATA sequence located between the Ϫ10 and Ϫ35 elements contributes to promoter activity. Therefore, to determine if additional motifs in the psbD BLRP confer promoter activity, we tested the influence of five point mutations in the sequences Boxes indicate native psbD BLRP sequences, and the nucleotides shown in boldface letters indicate sequences from the cloning vector. The same primer used for sequencing reactions was also used for the primer extension analyses shown to the right of each set of sequencing lanes.
THe arrows indicate transcript 5Ј termini revealed by this analysis. located between the AAG-box and the prokaryotic Ϫ10 element (see Fig. 4A, nt Switch). As shown in Fig. 4B, these substitutions did not alter transcription from the psbD BLRP, suggesting the absence of important sequence motifs between the AAG-box and the Ϫ10 promoter element (Fig. 4B, lane 6).
The AGF, which binds to the AAG-box, may stabilize and orient the RNA polymerase relative to the Ϫ10 element of the psbD BLRP. Therefore, spacing between the AAG-box and the Ϫ10 element may be important to maintain alignment of AGF and the RNA polymerase on the same face of the psbD BLRP. The AAG-box and the Ϫ10 motif are separated by 23 bp in the psbD BLRP. In contrast, most plastid Ϫ10 and Ϫ35 elements are separated by 18 bp (reviewed in Ref. 16). Therefore, the influence of altering the spacing between the AAG-box and the prokaryotic Ϫ10 element was investigated. Nucleotide deletions (5 and 10 bp) or insertions (3, 7, and 10 bp) were introduced between the AAG-box and the Ϫ10 element in plasmid pLRP140 (Fig. 4A). When a 5-bp deletion was introduced to reduce spacing between the AAG-box and the Ϫ10 element to 18 bp, transcription from the psbD BLRP was undetectable (Fig. 4B, lane 7). Deletion of 10 bp, which represents one helical turn, resulted in low but detectable levels of transcription (Fig.   4B, lane 8). Insertion of 3, 7, or 10 bp between the AAG-box and the Ϫ10 element also reduced transcription to very low levels (Fig. 4B, lanes 9 -11). These results indicate the importance of the 23-nucleotide spacing between the AAG-box and the Ϫ10 element for transcription from the psbD BLRP.
Further Analysis of Sequences in the AAG-box-The AAGbox, shown in Fig. 5A, was defined in past experiments as a 22-bp DNA region (Ϫ36 to Ϫ57) containing two motifs designated aaЈ and bbЈ (37). We have previously shown by point mutation analyses that the aaЈ motif is important for both AGF binding and transcription from the psbD BLRP (34). To test the importance of the bbЈ motif for transcription from the psbD BLRP, we introduced point mutations in this sequence (GAC-CTGACT) in plasmid pLRP140 (see Fig. 5A). Transcription analysis showed that mutation of GACC to GTAG inhibited transcription from the psbD BLRP (Fig. 5B, lane 1 versus lane  2 and lane 4 versus lane 5). Furthermore, mutations of GAC-CTGACT to GTAGTGTGA abolished transcription from the psbD BLRP (Fig. 5B, bbЈ mt, lanes 3 and 6). To determine whether the bbЈ sequences within the AAG-box also influence binding of AGF, gel retardation and competition binding experiments were carried out (Fig. 5C). As observed previously (34), AGF binds to radiolabeled pLRP140 in the absence of specific competitor DNA fragments (Fig. 5C, lane 1). The addition of unlabeled pLRP140 to the binding assays greatly reduces the amount of AGF gel shift complex (Fig. 5C, lane 2). As described previously, LRP140 DNA fragments containing modified aaЈ sequences (AAAGTAAG to AAATTCAT) do not compete well with native LRP140 (lane 3) (34). Modification of bbЈ sequences in pLRP140 (b mt and bbЈ mt) reduces the ability of the resulting DNAs to bind AGF to some extent (Fig. 5C, lanes 4 and 5), indicating that the bbЈ sequence contributes to AGF binding either directly or indirectly.

Delineation of a 53-bp Core psbD BLRP Promoter Domain-
The psbD BLRP is located approximately 570 bp upstream of the psbD translational start codon in cereals and even further upstream of the psbD open reading frame in dicots (37). In higher plants, a DNA region of approximately 130 bp surrounding the site of transcription initiation from the psbD BLRP is highly conserved (ϳ60%) relative to sequences more than 100 bp upstream of the promoter or sequences between the promoter and the psbD open reading frame (9%) (37). At least 25 bp of the conserved region extends downstream of the site of transcription initiation. In this study, we determined that deletion of sequences from Ϫ5 to ϩ64, relative to the site of transcription initiation, had no influence on transcription from the psbD BLRP in vitro. This result indicates that the conserved sequences downstream of the initiation site are probably not important for transcription. Previous analysis of changes in psbD transcription and RNA levels during leaf and chloroplast development indicated that psbD transcripts become more stable during light-mediated leaf maturation (8,13). Therefore, the conserved sequences immediately downstream from the site of transcription initiation, which are present in the 5Ј-untranslated region of transcripts produced from the psbD BLRP, may be important for RNA stability.
The 100-bp DNA region immediately upstream of the psbD BLRP initiation site contains several stretches of sequence that are conserved among psbD genes from higher plants (37). Deletion of sequences from Ϫ107 to Ϫ55 in the tobacco psbD BLRP reduced transcription activity in vivo ϳ5-fold without altering light-stimulated transcription following dark adaptation of plants (40). In barley, this region of the psbD BLRP specifically binds a protein complex (PGTF) present in chloroplasts (34). In the current study, however, deletion of sequences upstream of Ϫ57 in the psbD BLRP had minimal effect on in vitro transcription. This suggests that this region of the psbD BLRP and the PGTF complex that binds in this region are not modulating transcription from the psbD BLRP in vitro. Mutation of sequences immediately downstream of Ϫ57 (34) or upstream of Ϫ5 (Fig. 5) reduce transcription from the psbD BLRP. These experiments define a 53-bp region that is required for transcription from the psbD BLRP in vitro.
Transcription from the psbD BLRP Requires a Prokaryotic Ϫ10 Element but Not a Ϫ35 Promoter Element or the psbA TATATA Element-The psbD BLRP contains the sequence TATTCT, located between Ϫ7 and Ϫ12, which resembles a prokaryotic Ϫ10 promoter element. Mutation of this sequence to AATTCA reduced transcription from the psbD BLRP to very low levels. Similarly, mutation of Ϫ10 sequences found in the psbA (TATACT to AATACA) and rbcL (TACAAT to AACAAA) promoters rendered these promoters inactive. In Escherichia coli, Ϫ10 promoter elements are recognized via interaction with -factors that are associated with the RNAP (reviewed in Refs. [52][53][54]. These results are consistent with in vitro transcription of the psbD BLRP by a chloroplast RNAP containing a -like subunit that interacts with the Ϫ10 promoter element (29,31,(55)(56)(57). Transcription from mustard psbA is stimulated by a TATATA sequence located between the Ϫ10 and Ϫ35 promoter elements (46). The TATATA sequence might be involved in the recruitment of RNA polymerase or in the isomerization from the "closed" to "open" complex formation (Refs. 58 and 59; reviewed in Refs. 60 and 61). Moreover, in mustard, this sequence may allow transcription in dark-grown plants that is not dependent on a Ϫ35 element from the psbA promoter (31,46). Mutation of a similar sequence present in the barley psbA promoter decreased transcription in plastid extracts from dark-grown and illuminated plants (Fig. 3). In contrast, the psbD BLRP lacks the TATATA sequence, and mutation of sequences located between Ϫ10 and Ϫ35 in the psbD BLRP had little influence on transcription activity.
The chloroplast-encoded RNAP's ability to transcribe rbcL and psbA depends on a prokaryotic Ϫ35 promoter element (Figs. 3 and 4) (reviewed in Refs. 16 and 49). In contrast, mutation of the Ϫ35 sequence in the psbD BLRP had little effect on transcription in vitro (Fig. 5). The function of the Ϫ35 sequence in the psbD BLRP appears to be replaced by the action of AGF, an activating complex that binds immediately upstream of the Ϫ35 sequence (Ref. 34; see below).
Two Different Sequences in the AAG-box Are Involved in psbD BLRP Transcription-The sequence from Ϫ36 to Ϫ64 in the psbD BLRP was previously reported to be required for transcription from the psbD BLRP in vitro (34). In the current study, this region was further truncated to Ϫ57 without loss of activity. The corresponding sequence in the tobacco psbD BLRP was also found to be important for activity in vivo (40). The region from Ϫ36 to Ϫ57, termed the AAG-box, was previously reported to contain two conserved motifs (aaЈ and bbЈ) (37). A protein complex, designated AGF, was found to specifically interact with sequences within the AAG-box. Footprint analysis indicated that AGF binding protected sequences from at least Ϫ40 to Ϫ63 (34). In a previous study, site-directed mutagenesis of the aaЈ sequence (AAAGTAAGT to AAATTCAT) caused loss of AGF binding and eliminated transcription from the psbD BLRP (34). In the current study, site-directed mutagenesis of the bbЈ sequence located immediately downstream from the aaЈ motif and upstream of Ϫ35 caused a reduction in control binding assay in the absence of competitor DNA is shown in lane  1 (No Comp.). The arrow designates the AGF-binding complex consistent with previous analysis (34). The migration of free probe is indicated. The asterisk indicates a signal produced from plastid extracts in the absence of template. DNA size markers in base pairs are indicated to the right. C, competition binding assays carried out using radiolabeled LRP136, which contains the entire psbD-psbC blue light-responsive promoter, using high salt extracts from plastids isolated from 4.5-day-old dark-grown barley. Competiton reactions were carried out with 200 ng of competitor DNA in the presence of 1 g of poly(dI-dC)⅐(dI-dC). Competitor DNAs included the native sequence (pLRP140; labeled 140), mt2 (LRP140 containing a mutation in the aaЈ sequence) (34), b mt, or bbЈ mt DNAs shown in A. A transcription as well as a reduction in the ability of DNA in this region to bind to AGF (Fig. 5). These results suggest that proteins in AGF interact with the bbЈ sequence. It is also possible that some other currently undetected protein binds to the bbЈ sequence and that this modifies AGF binding. In tobacco, proteins also bind specifically to the bbЈ sequence (40). Unfortunately, the relationship between the barley and tobacco AAG-box binding complexes could not be established.
Model for AGF Activation of the psbD BLRP-A model of the barley psbD BLRP is shown in Fig. 6 along with diagrams of the rbcL and psbA promoters. All three genes are shown being transcribed by the chloroplast-encoded RNAP with an associated -factor. This is consistent with several lines of evidence. First, light-induced transcription from the psbD BLRP in vivo is inhibited if plants are pretreated with tagetitoxin (13). The chloroplast-encoded RNAP and E. coli RNAP are sensitive to tagetitoxin, whereas the chloroplast-localized, nucleus-encoded RNAP and the homologous bacteriophage RNA polymerases, T7 or SP6, are not inhibited by tagetitoxin (62,63). Second, plants that lack the chloroplast-encoded RNAPs do not accumulate transcripts from the psbD BLRP (or from rbcL, psbA), although they accumulate transcripts from many genes involved in transcription and translation that lack prokaryotic Ϫ10 and Ϫ35 promoter elements (22,64). Third, mutation of sequences surrounding the psbD BLRP site of transcription initiation (*) from TTCTGATATAT*AAAT to TTCTGAGGAT-C*CCCC had no influence on transcription in vitro (Figs. 1 and  2). The nucleus-encoded chloroplast RNAP has been proposed to use a promoter sequence located in the 10 bases immediately adjacent to the site of transcription initiation (64). Based on comparative alignments, a rather variable promoter consensus sequence, ATAGAAT(A/G)AA, has been proposed for this polymerase (24,64). This sequence is somewhat different from both the native and mutated psbD BLRP promoters that are active in vitro. Fourth, mutation of the prokaryotic Ϫ10 element, located between Ϫ7 and Ϫ12, dramatically reduced transcription from this promoter. Finally, the chloroplast-encoded RNAP preferentially transcribes genes encoding proteins involved in photosynthesis; therefore, transcription from the psbD BLRP is consistent with this tendency. However, further biochemical analysis of the nucleus-encoded RNAP will be needed to definitively eliminate a role for this RNAP in psbD BLRP transcription.
The RNAPs in Fig. 6 are shown associated with a generic -factor. However, there are several reasons to think that the -factor involved in transcription of the psbD promoter may be different from -factors involved in transcribing rbcL and psbA. First, in the case of the rbcL and psbA promoters, -factors are likely to interact with both Ϫ10 and Ϫ35 promoter elements, based on analysis of bacterial -factor binding (reviewed in Refs. 52 and 65). An additional interaction may occur between the -factor and the TATATA sequence in the psbA promoter. In contrast, the psbD BLRP lacks functional Ϫ35 and TATATA elements, and the sequence of its Ϫ10 element differs from those of rbcL and psbA. Second, the psbD AAG-box did not activate transcription when fused upstream of a derivative of the rbcL promoter shown in Fig. 3, which lacks an active Ϫ35 element (data not shown). This could mean that AGF interacts with an RNA polymerase containing a -factor that is incompatible with the rbcL promoter. Third, utilization of a different -factor for transcription of the psbD BLRP would allow blue light-specific regulation of this promoter via the -factor. Recently, genes encoding three chloroplast -factors have been cloned (29,30). Moreover, the expression of at least one -factor gene is regulated by light (56,66), and previous work showed that these factors are the target of light-mediated regulation of chloroplast transcription (31).
The function of the Ϫ35 promoter element in the psbD BLRP is likely to be replaced by an activating complex bound to the AAG-box (Fig. 6, AGF/BBЈ). The AAG-box contains two binding domains, aaЈ and bbЈ, which bind AGF. The AGF, unlike -factors, binds to DNA in the absence of the RNAP (34). A subunit of AGF or perhaps a separate protein, noted in Fig. 6 as BBЈ, binds specifically to the bbЈ motif. The AGF/BBЈ could activate the psbD BLRP by recruiting the RNA polymerase to the psbD BLRP, by stabilizing the binding of the RNAP to the BLRP, or by changing RNAP recognition of the Ϫ10 element, thus promoting transcription (reviewed in Ref. 67).
The structure of the psbD BLRP shown in Fig. 6 resembles a class of bacterial promoters that use activating proteins to stimulate transcription (reviewed in Refs. 52 and 65). The activating sequences in one class of these promoters (type I; i.e. cAMP receptor protein binding site in lacP1) can be moved various distances upstream of the promoter (68). In type II promoters such as galP1, the site of activator binding must be immediately upstream of Ϫ35 (68). In both cases, the ␣-subunit of RNAP interacts with the activating complex (Refs. 68 and 69; reviewed in Ref. 70), although in different ways (71). In this regard, the psbD BLRP is similar to a type II bacterial promoter. The addition of 3, 7, or 10 bp between the Ϫ10 element and the AAG-box dramatically inhibited transcription, indicating that the AGF factor needs to be approximately 23 bp from the Ϫ10 element. Moving the AAG-box closer to the Ϫ10 element by removal of five nucleotides between the Ϫ10 and AAG-box also inhibited transcription. However, constructs with deletion of 10 bp still showed a low level of activity. Deletion of 10 bp, or one helical turn, would keep the AAG-box and the Ϫ10 element in the same relative orientation along the DNA helix. Therefore, a low level of transcription from this FIG. 6. Models of transcription complexes associated with the psbD BLRP, rbcL and psbA promoters. Arrows (ϩ1) indicate the site and direction of transcription initiation. Important transcription cis-elements (-10, -35, TATA, and aaЈ/bbЈ) are boxed and the sequences and spacing between elements is indicated. A chloroplast RNAP and an associated sigma factor is shown interacting with each promoter. In addition, the AGF/BBЈ complex, which binds to the AAG-box sequences aaЈ/bbЈ, is shown interacting with the RNAP to promote transcription from the psbD BLRP. template is possible, although packing of the RNAP and AGF on the template must be tight.
Regulation of the psbD BLRP-Illumination of 7.5-day-old, dark-grown barley with white light caused a 10-fold increase in transcription from the psbD BLRP and a 4-fold increase in transcription from rbcL in vivo (72). Surprisingly, in vitro transcription of the 53-bp psbD BLRP in plastid extracts from 7.5-day-old, dark-grown plants that had been illuminated for 16 h, was approximately 6.5-fold higher than in extracts of dark-grown plants (Fig. 4). Transcription from the rbcL promoter was also approximately 2-fold greater in extracts from illuminated plants (Fig. 3). This suggests that light-induced modifications that activate transcription in vivo are retained in vitro. Light could induce the accumulation of a transcription factor and/or cause modification of the RNAP, a -factor, or the AGF during the illumination period. Inhibitor studies have implicated the involvement of protein kinases and phosphatases in blue light modulation of transcription from the psbD BLRP (73). Future experiments will be directed toward identification of the potential targets of these protein kinases/phosphatases and an understanding of their role in blue light modulation of the psbD BLRP.