Continuous Primary Sequence Requirements in the 18-Nucleotide Promoter of Dicot Plant Mitochondria*

The nucleotide requirements of mitochondrial promoters of dicot plants were studied in detail in a pea in vitro transcription system. Deletions in the 5′ regions of three different transcription initiation sites from pea, soybean, andOenothera identified a crucial AT-rich sequence element (AT-Box) comprising nucleotide positions −14 to −9 relative to the first transcribed nucleotide. Transversion of the AT-Box sequence to comple- mentary nucleotide identities results in an almost complete loss of promoter activity, suggesting that primary structure rather than a simple accumulation of adenines and thymidines in this region is essential for promoter activity. This promoter segment thus appears to be involved in sequence specific binding of a respective protein factor(s) rather than merely loosening and melting the DNA helix during or for an initiation event. Manipulation of nucleotide identities in the 3′ portion of the pea atp9 promoter and the respective 3′-flanking region revealed that essential sequences extend to positions +3/+4 beyond this transcription start site. Efficient transcription initiation at an 18-base pair promoter sequence ranging from nucleotide positions −14 to +4 integrated into different sequence contexts shows this element to be sufficient for autonomous promoter function independent of surrounding sequences.

The nucleotide requirements of mitochondrial promoters of dicot plants were studied in detail in a pea in vitro transcription system. Deletions in the 5 regions of three different transcription initiation sites from pea, soybean, and Oenothera identified a crucial AT-rich sequence element (AT-Box) comprising nucleotide positions ؊14 to ؊9 relative to the first transcribed nucleotide. Transversion of the AT-Box sequence to complementary nucleotide identities results in an almost complete loss of promoter activity, suggesting that primary structure rather than a simple accumulation of adenines and thymidines in this region is essential for promoter activity. This promoter segment thus appears to be involved in sequence specific binding of a respective protein factor(s) rather than merely loosening and melting the DNA helix during or for an initiation event. Manipulation of nucleotide identities in the 3 portion of the pea atp9 promoter and the respective 3-flanking region revealed that essential sequences extend to positions ؉3/؉4 beyond this transcription start site. Efficient transcription initiation at an 18-base pair promoter sequence ranging from nucleotide positions ؊14 to ؉4 integrated into different sequence contexts shows this element to be sufficient for autonomous promoter function independent of surrounding sequences.
The established endosymbiont hypothesis describes mitochondria and chloroplasts as remnants of bacteria-like progenitors, which penetrated the original eukaryote 1.5 and 1 billion years ago (1). The prokaryotic origin of the organellar DNAs implies that cis-elements engaged in the expression of the organellar genetic information might be similar to the respective control structures of modern bacteria. Such a similarity is indeed found in chloroplasts, where several components of the genetic system show a close structural and functional relationship to their prokaryotic counterparts (2). Many chloroplast promoters for instance resemble the Ϫ10/Ϫ35-type promoters of contemporary prokaryotes. These promoters serve as ciselements for a largely chloroplast encoded eubacteria-like RNA polymerase. Another, nuclear encoded polymerase initiates transcription at a second less well described class of promoters (3)(4)(5)(6). This latter enzyme is of the bacteriophage-like single subunit type, a class of polymerases that also includes the RNA polymerases in mitochondria of animals, fungi, and plants (7)(8)(9). These enzymes bind to promoters in close spatial relationship to transcription initiation sites.
In mitochondria of Saccharomyces cerevisiae a 9-base-pair-long highly conserved cis-element is found at each of the about 20 transcription initiation sites. This nonanucleotide motif autonomously supports efficient transcription initiation in vitro, with enhanced promoter activity depending on the presence of a purine at position ϩ2 and a pyrimidine at ϩ3. In Xenopus laevis mitochondria an eight-nucleotide-long cis-element drives bi-directional transcription initiation at two different locations. A 15-bp 1 consensus motif is sufficient for basal promoter activity in mammalian mitochondria, while maximal promoter function depends on additional upstream sequence elements (10,11).
In vitro capping analyses have identified conserved sequence elements also at mitochondrial transcription initiation sites in plants. These investigations revealed the tetranucleotide 5Ј-CRTA-3Ј to be the only common motif found at initiation sites of both monocot and dicot plant species (12)(13)(14)(15)(16). Functional studies of the maize atp1 promoter in a homologous in vitro transcription system confirmed the significance of some of the conserved sequence features (i.e. 5Ј-CRTA-3Ј), while other segments appear to be not required at least for basic promoter function in vitro (e.g. G(AT) 3-4 ; Refs. 16 and 17). Further investigations revealed also differences in the basic structures and requirements between individual promoters. While an A-rich upstream domain contributes significantly to the maize atp1 promoter function, such a domain appears not to be required for the cox3 promoter activity in the same species (18,19).
Mitochondrial promoters in dicot plant species display a somewhat larger conserved sequence element compared with those found in monocots. Inspection of sequences adjacent to transcription initiation sites of several bona fide dicot promoters revealed a conserved nonanucleotide motif (5Ј (Ϫ7)-CRTAA-GAGA (ϩ2) -3Ј, CNM), which covers the first two transcribed nucleotides (20). Functional studies, however, showed that this motif per se is not sufficient for the function of the pea atp9 promoter in vitro and that additional sequences are required (21). We have now determined these additional sequence requirements and report that essential promoter sequences require specific nucleotide identities within an 18-nucleotide region between positions Ϫ14 and ϩ4 relative to the transcription initiation site.

EXPERIMENTAL PROCEDURES
Template Construction-5Ј progressively deleted fragments of pea atp9, soybean atp9, and Oenothera rpl5 promoter region were amplified by PCR using Pfu polymerase in buffer supplied by the manufacturer (Stratagene). PCR conditions, primer pairs, and DNA templates were as follows. Pea atp9 promoter region (patp9): dpϩ355 and one of the oligonucleotides dp-16, dp-14, dp-12, dp-10, and dp-7 using patp9SK630 as template (21); soybean atp9 promoter region: Rev and one of the oligonucleotides GM-D-7, GM-D-10, GM-D-12, GM-D-14, and GM-D-16 using clone satp9XR482 as DNA template (21); Oenothera rpl5 pro-* This work was supported by grants from the Deutsche Forschungsgemeinschaft and a Landesforschungsschwerpunkt Baden-Wü rttemberg. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For substitutions in the 3Ј-flanking region of the pea atp9 conserved nonanucleotide motif inverse PCR was performed with oligonucleotides LS-5-1 and LS-3-1 for clone LS-1 and LS-5-2 and LS-3-2 for mutant LS-2 on patp9SK630 and patp9SK630 ⌬-172 clones as templates, respectively. Partial digestion by PstI resulted in variations of the inserted sequences. Due to cloning artifacts during construction of LS-2 an XbaI site was used for linearization instead of KpnI with construct LS-1.
Inverse PCRs were exclusively performed with Klentaq polymerase (CLONTECH) in a buffer supplied by the manufacturer with 35 cycles under the following conditions: 1 min 94°C, 1 min 46°C (TM mutants) and 50°C (LS mutants), respectively, and 4 min at 68°C. After PCR the template DNA was removed by digestion with DpnI, the DNAs were separated on agarose gels and the respective fragments were ligated at 15°C for at least 16 h.
All DNA templates used in the experiments described in this report are controlled by restriction digests and complete sequence analysis. Sequences of all oligonucleotides used are available on request.
About 18 pmol of each MP oligonucleotide (MP sense: 5Ј-ATAATAG-CATAAGAGAAG-3Ј and the respective antisense oligonucleotide) were denatured for 30 s at 80°C, incubated for 20 min at 28°C, and slowly adapted to room temperature in a total volume of 10 l. Prior to ligation into the SmaI and EcoRV sites in pBluescript and the SnaBI site of a clone containing the pea rps14/cob spacer region, ϳ9 pmol of doublestranded MP oligonucleotides were concatemerized in the presence of 12% (v/v) polyethylene glycol, 50 nmol ATP, 1 mM dithiothreitol, and 10 units of T4 DNA ligase. 5 l of the concatemerized MP oligonucleotides were then cloned into pBluescript vectors and a rps14/cob clone in the presence of 0.05 mM ATP, 1 mM dithiothreitol, 11% (v/v) polyethylene glycol, and 10 units of T4 DNA ligase in a total volume of 15 l.
In Vitro Transcription Assays-Enrichment of pea mitochondrial lysates and in vitro transcription reactions were performed as described previously (21). Between 23 and 47 g of lysate were used in standard assays with 100 ng of template DNA quantified both photometrically and on agarose gels.
Miscellaneous Methods-Standard methods in molecular biology were performed following Sambrook et al. (23). DNA sequencing was done with Thermo Sequenase fluorescent labeling kits (Amersham Pharmacia Biotech). Cy5-labeled sequencing products were detected on line with an ALF express sequencer (Amersham Pharmacia Biotech).

In Vitro Analysis of the Upstream Extent of the Pea atp9
Promoter-Initial studies of the mitochondrial pea atp9 promoter in a homologous in vitro transcription system had confirmed the functional importance of the nonanucleotide between nucleotides Ϫ7 and ϩ2, which is conserved between several promoters. In addition the analysis of clones containing 5Ј-deleted promoter regions suggested that crucial sequence information for the function of this promoter is located between nucleotide positions Ϫ25 and Ϫ8 relative to the 5Ј end of the nascent transcript (21). To identify the functional requirements within this region, a series of deletion clones was constructed and assayed in the in vitro transcription system. While constructs with deletions upstream of positions Ϫ25, Ϫ21, and Ϫ16 contain full promoter function, the initiation activity is significantly reduced in construct ⌬-12 and completely lost in mutant ⌬-7 (data not shown). For further resolution of the nucleotide requirements in the comparatively AT-rich region between positions Ϫ16 and Ϫ7, the mitochondrial sequences were deleted in smaller steps (Fig. 1, upper part). While promoter function is not impaired in constructs ⌬-16 and ⌬-14, activity is entirely lost in deletion clones ⌬-10 and ⌬-7. Partial transcription initiation activity is again observed with clone ⌬-12. Independent reruns of these experiments, which were calibrated by the addition of T7 RNA polymerase-generated transcripts to the in vitro transcription assays, confirmed that the reduced amount of run-off products originated from decreased intrinsic promoter activity of this template and not from a selective loss of in vitro generated RNAs during the assays (Fig. 1, lower part, T7(161 Nt)). These functional analyses demonstrate that sequences extending up to nucleotide position Ϫ14 are sufficient and necessary to support transcription initiation at the pea atp9 promoter in vitro.
Sequences Extending to Nucleotide Position Ϫ14 Are Generally Sufficient for Promoter Function in Vitro-To gain more information about the general significance of 5Ј sequences in other dicot promoters, analogous deletion studies were carried out with the soybean atp9 and the Oenothera rpl5 promoters, respectively. In the assays with soybean constructs no functionally required sequences are lost upon deletion upstream of position ⌬-14, but promoter activity is seriously reduced in clone ⌬-12 (Fig. 2). In Oenothera rpl5 deletion constructs no significant differences in transcription efficiencies were found between ⌬-17, ⌬-14, and ⌬-13. Promoter activity is, however, completely abolished in deletions ⌬-10 and ⌬-7 in both soybean atp9 and Oenothera rpl5 promoters. In summary these results support the conclusion that the promoter portions downstream of nucleotide position Ϫ14 are generally sufficient and also necessary to drive transcription at mitochondrial promoters of dicot plants. This region with a high content of adenines and thymidines, which is already conspicuous in the sequence alignment of several dicot promoters, has been accordingly designated AT-Box (21).
Transversion Mutagenesis of the AT-Box-The high content in adenosines and thymidines in the above described 5Ј regions between nucleotides Ϫ14 and Ϫ8 of the mitochondrial promoters in dicot plants suggests a function of these sequences in loosening or melting of the DNA strands during the initiation process. To determine whether this concentration of A and T nucleotides per se or whether sequence specificity is important, adenosine and thymidine residues in this particular region of the pea atp9 promoter were transverted to the respective complementary nucleotides (Fig. 3, upper part). The melting temperatures of the mutated and the wild type promoter regions were calculated to differ by only 0.6°C, and thus comparable initiation efficiencies are expected if the function of this segment is confined to access strand opening of the DNA. However, in vitro transcription assays with the mutated and wild type templates revealed that initiation is almost completely abolished by the transverted nucleotides (Fig. 3, lower part). Semiquantitatively repeated assays confirmed the significant reduction of the promoter activity. The faint signal is only visible upon long exposures, which indicates the residual promoter function to be less than 1% of the wild type activity. This observation suggests that sequences rich in adenosines and thymidines per se do not support promoter function and that this promoter segment contributes to the sequence-specific interaction(s) between the promoter and the transcriptional apparatus in dicot plant mitochondria.
Insertions of 1, 3, 5, or 11 guanosine nucleotides between the AT-Box and CNM also almost completely abolished promoter function. Since promoter activity is not restored even when the distance between the two parts corresponds to a complete helix turn, the AT-Box and CNM appear to be interdependent in function (data not shown).
Nucleotides Downstream of the Transcription Start Site Contribute to Promoter Function-Analogous to the definition of the 5Ј border of different dicot promoters, their 3Ј extent was investigated with the pea atp9 transcription initiation site. Two 8-bp segments ranging from ϩ3 to ϩ10 (LS-1) and from ϩ11 to ϩ18 (LS-2) were replaced by 15-and 14-bp insertions of alien sequences, respectively. In vitro transcription assays with both templates indicate that the exchanged sequences are not important for promoter activity at least in vitro (data not shown). To extend this analysis further upstream into the conserved nonanucleotide motif, in three clones segments between nucleotides Ϫ3 to ϩ2 (TM-1), ϩ3 to ϩ7 (TM-2), and ϩ8 to ϩ12 (TM-3) were transverted to the respective complementary sequences (Fig. 4A, upper part). The correct transcripts of 355 nucleotides obtained with construct TM-3 confirmed that the region downstream of position ϩ8 does not participate in transcription initiation in vitro. The ability to initiate transcription is however completely lost in mutants TM-1 and TM-2 (Fig. 4A, lower part). While the loss of promoter activity in TM-1 can be explained by the disruption of the CNM, the inactivity of mutant TM-2 contradicts the relatively normal promoter activity observed with construct LS-1. To clarify these apparently inconsistent results, an additional mutant was generated, in which nucleotides ϩ5 to ϩ7 were transverted (Fig. 4B, TM-2Ј, upper part). While in vitro transcription assays with constructs LS-1 and TM-2 confirmed the previous results, mutant TM-2Ј does retain promoter activity. In summary these results show that sequences extending into transcribed regions up to nucleotide position ϩ4 are sufficient and at least partially required to drive plant mitochondrial transcription in vitro. 18 Nucleotides Are Sufficient to Direct Transcription Initiation Irrespective of Surrounding Sequences-The above results of the functional analysis of the 5Ј and 3Ј regions of the pea atp9 promoter imply that transcription should be initiated at an 18-bp sequence segment (Ϫ14 to ϩ4) irrespective of 5Ј-and 3Ј-flanking regions. A double-stranded oligonucleotide (MP) representing such a mitochondrial promoter was cloned into three different sequence contexts and tested for transcription initiation (Fig. 5). The MP oligonucleotides integrated into the SmaI and EcoRV restriction sites of the pBluescript multiple cloning sites show the same full promoter activity as the respective control template with this sequence in its wild type context (pS and pE, Fig. 5, A and B).
Two 18-mer promoter sequences inserted in tandem into the EcoRV restriction site yield two transcription products differing by 18 nucleotides in length (pEx2, Fig. 5B). No specific transcription product is observed with the pBluescript vector alone clearly correlating transcription initiation with the integrated MP oligonucleotide(s). Insertion of the MP oligonucleotide into a SnaBI site within the pea rps14/cob spacer region confers transcription initiation to this region where no promoter is observed in vivo and in vitro in the wild type sequence (data not shown). These results conclusively demonstrate that the sequence from Ϫ14 to ϩ4 represents a fully functional di-cot plant mitochondrial promoter autonomous of adjacent sequences.

The Mitochondrial Promoter in Dicot Plants Is a Single Entity of 18
Nucleotides-The functional analyses of several promoters by deletions and transversions 5Ј and 3Ј to the transcription start sites show that 18 nucleotides from Ϫ14 to ϩ4 are essential and sufficient for transcription in vitro. Insertion of the 18mer DNA segment confirms its fully competent promoter function independent of 5Ј-and 3Ј-flanking sequences. This comprehensive analysis thus identifies extension and structure of such a promoter.
To evaluate the experimental results, the respective regions of eleven promoters from dicot plants were reinvestigated for conserved structures (Fig. 6). Only transcription initiation sites were included in this comparison, which have been unambiguously identified as such by primer extension (in vivo) and in vitro capping and/or in vitro transcription analysis (in vitro) (12)(13)(14)24). 2 Most striking features within this 18-mer promoter sequence are the CNM (Ϫ7 to ϩ2) and the upstream AT-Box (Ϫ14 to Ϫ8). Although the apparently noncanonic nucleotide identities separate these two motifs, the in vitro analyses do not support individual functions of these different promoter portions. The promoter sequence rather appears to function as a single element, in which most nucleotides are involved in sequence-specific interaction with the RNA polymerase and/or additional transcription factor(s).
Requirement of the Primary Structure of the AT-Box for Promoter Function-Deletions in the 5Ј portion of three different promoters from three dicot plant species consistently show that functional sequences maximally extend to position Ϫ14. Transcription initiation at the Oenothera rpl5 construct ⌬-13 occurs with an efficiency equal to clone ⌬-14. This experimental result and the observation that the adenosine at position Ϫ14 is less conserved between different promoters (36%) suggest that this nucleotide position may not be crucial for the activity of all of these promoters (Figs. 1, 2, and 6).
Speculations that the AT-Box may only be necessary to facilitate melting of the DNA strands are rendered highly unlikely by the almost complete loss of activity in the transverted AT-Box mutant. This sequence dependence is reflected in the high conservation at positions Ϫ13 (73% A), Ϫ12 (82% A), Ϫ11 (91% A), and Ϫ10 (73% T) within this promoter segment. These observations strongly support this part of the promoter to be involved in sequence-specific interaction with the transcription apparatus, which does of course not exclude a parallel positive influence of these AT-rich sequences on melting of the DNA strands.
The increased promoter activity in constructs ⌬-14 relative to ⌬-16, which is reproducibly observed in independent experiments with both pea and soybean atp9 constructs, suggests that sequences upstream of position Ϫ14 may somehow modulate promoter strength. Thus substitution of nucleotide identities with negative impact on promoter activity could result in a positive effect, a phenomenon, which has to be investigated in detail by site-directed mutagenesis at these particular nucleotides. Character sizes in the consensus reflect the frequency with which the nucleotide identity is found at the given position. Promoters compared in this analysis are located upstream from the following genes: atp9, 18 S rRNA, and rpl5 genes from pea; atp1, cox2, rpl5, trnF, 18 S rRNA from Oenothera; atp9 and rb/c from soybean; and the 18 S rRNA gene of potato (12-14, 21, 24). 2 Adenosines in positions Ϫ12 to Ϫ10 have been observed to also influence the activity of the maize atp1 promoter. Substitution of the Ϫ12 A by a T or C considerably reduced the amount of run-off transcripts (between 30 and 50%) of the wild type level (18). These similarities between the 5Ј regions of the maize atp1 and the dicot promoters investigated here may indicate analogous promoter structures and recognition mechanisms. In the maize cox3 promoter the respective segment was, however, found to have only a minor role in transcription initiation in vitro suggesting major differences of sequence requirements between individual promoters in this plant species (19).
Nucleotides Downstream of the 5Ј-CRTA-3Ј Are Conserved and Essential for the Function of Promoters in Dicot Plants-Alignment and sequence comparisons of transcription initiation sites determined by in vitro capping studies identified the 5Ј-CRTA-3Ј tetranucleotide to be conserved between monocot and dicot plant species. Consequently this motif was assigned an important role in promoter function (15,16,25). Various in vitro promoter studies, however, showed that deviations are accepted in the first two positions of this tetranucleotide in both plant subdivisions (18,26). 2 The small size of the motif and the relaxed sequence constraints indicate that other nucleotides must be essential for promoter function. In addition to the importance of the upstream AT-Box, the nucleotides downstream of the 5Ј-CRTA-3Ј contribute significantly to promoter function. This is highlighted by the complete lack of initiation in the transversion mutant TM-1, in which the second half of the nonanucleotide sequence downstream of the 5Ј-CRTA-3Ј motif is altered. The theoretical consensus deduced in the sequence comparison shows that this promoter segment contains three nucleotide positions at Ϫ2 (G), ϩ1 (G), and ϩ2 (A), which are 100% conserved in all compared dicot promoters. Furthermore, two adenosine residues are found at Ϫ3 and Ϫ1 with a frequency of 82%. This indicates that the 5Ј-CRTA-3Ј motif, although present in almost all mitochondrial promoters in seed plants, only represents a small fraction of the nucleotides contributing more or less equally to the promoter function.
Purine Nucleotides Downstream of the Transcription Start Site Contribute to Promoter Function-The identification of the nonanucleotide motif had already shown that the conserved promoter core extends beyond the transcription start site. Nucleotide transversions now show that the transcribed promoter portion actually extends even further downstream, as far as nucleotide positions ϩ3 and/or ϩ4. Here the actual requirements on nucleotide identity may be relaxed, since not all substitutions downstream of position ϩ2 interfere with successful transcription initiation (Fig. 5). The differing results obtained with constructs LS-1 and TM-2 may be due to the particular nucleotide identities inserted at positions ϩ3 and ϩ4, these being a guanosine at ϩ3 in LS-1 and a thymidine residue in the respective position in TM-2. Although guanosine or thymidine residues are found in this position in some other promoters, at least one purine is found conserved at ϩ3 or ϩ4 in all wild type promoter sequences. Since TM-2 contains only pyrimidines in these positions, the loss of promoter activity of this construct could be due to the total absence of purine here. The precise requirements of nucleotide identities in these positions have to be determined by site-directed mutagenesis.
Implications of the Refined Mitochondrial Promoter Structure in Dicot Plant for the Interaction with the Transcription Apparatus-The analysis of the pea atp9 transcription initiation site now shows that a mitochondrial promoter in a dicot plant covers an 18-bp sequence, which corresponds to about two helix turns. This promoter appears to act in its entirety as recognition site for components of the transcription apparatus. This may include one or two proteins, which could, similar to transcription factors in mammalian, yeast, and Xenopus mitochondria, direct a promoter-specific attachment of the RNA polymerase (10,11,27,28). Two proteins that bind to this promoter have recently been enriched; however their direct participation in the transcription initiation process remains to be shown (29). All involved proteins must be in very close proximity, since contact is made with the 18-bp promoter sequence, which is rather short to accommodate attachment sites for several different proteins, particularly considering that the RNA polymerase itself already has a molecular mass of more than 100 kDa (7,30).
The mode by which the transcriptional apparatus identifies a promoter sequence is at present unclear, but can be refined with the results presented here. The equal efficiencies with which transcription is initiated at the tandemly repeated promoters exclude an access of the proteins by scanning the DNA with a 5Ј to 3Ј processivity, since such an access would result in a preferential transcription initiation at the first promoter of this construct (Fig. 5). Thus, promoter recognition in plant mitochondria rather appears to involve a homing approach of one or more of the participating proteins.
The sequence requirements of mitochondrial promoters in dicot plants defined here now provide a base toward the identification of the components involved in transcription initiation. Furthermore, this refined promoter sequence will allow rapid identification of other mitochondrial promoters in this plant subdivision, e.g. in the complete sequence of the Arabidopsis mitochondrial genome (22,31).