INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In nearly all eukaryotic cells the core enzyme for mitochondrial RNA polymerase is a single-subunit protein homologous to those of bacteriophages (Cermakian et al., 1996; Hedtke et al., 1997; Gray and Lang, 1998). The evolutionary relation to bacteriophage RNA polymerases is evident, but the origin of these enzymes remains so far unclear (Cermakian et al., 1997; Lang et al., 1997). The single-subunit enzymes probably represent very ancient RNA polymerases that therefore exhibit no homologies to the eukaryotic multisubunit RNA polymerases of the nucleus (Sousa, 1996), despite of a comparable enzyme mechanism for transcription (Delarue et al., 1990; Sousa, 1996, Temiakov et al., 2000).

The mitochondrial enzymes are encoded in the nucleus, transcribed in the cytosol, and imported into the organelles (Greenleaf et al., 1986; Masters et al., 1987; Gray and Lang, 1998). Lately, bacteriophage-type single-subunit mitochondrial core enzymes have been characterized from humans (Tiranti et al., 1997) to higher plants (Hedtke et al., 1997; Weihe et al., 1997; Young et al.,1998; Chang et al., 1999; Hess and Börner, 1999). In the plant Arabidopsis thaliana three gene copies for organellar RNA polymerases could be identified (Börner et al., 1999; Hedtke et al., 2000). Two of them encode proteins of either plastid or mitochondrial location (Hedtke et al., 1997; Maliga, 1998), and the third encodes an enzyme that is imported into both organelles (Hedtke et al., 2000).

The single-subunit SP6/T7-type RNA polymerases of bacteriophages are characterized by a set of highly conserved domains that are supposed to be essential for catalytic functions in the process of transcription (Bonner et al., 1992, 1994a, 1994b; Sousa et al., 1992, 1993; Temiakov et al., 2000). Homologous domains are also present in the mitochondrial core enzymes, but these enzymes contain additional highly divergent amino- and carboxyl-terminal extensions (Masters et al., 1987; Cermakian et al., 1997; Hess and Börner, 1999). The yeast enzyme Rpo41p is characterized by an especially long amino-terminal extension of ~400 amino acids (Masters et al., 1987). Recently, it was shown that this extension is indispensable for stable maintenance of mitochondrial genomes (Wang and Shadel, 1999). Remarkably, the amino terminal domain could complement a respective deletion mutant of Rpo41p even in trans (Wang and Shadel, 1999).

Today, the yeast enzyme Rpo41p represents the best model for the basic functions of mitochondrial RNA polymerases (Kelly et al., 1986; Masters et al., 1987; Lisowsky and Michaelis, 1989; Lisowsky et al., 1996; Wang and Shadel, 1999). In contrast to the bacteriophage enzymes yeast mitochondrial RNA polymerase is dependent on the specificity factor Mtf1p (Schinkel et al., 1987; Lisowsky and Michaelis, 1988; Jang and Jaehning, 1991, Shadel and Clayton, 1995; Jan et al., 1999) for initiation of transcription at the conserved promoter sequences (Osinga et al., 1982; Schinkel et al., 1987; Xu and Clayton, 1992). In vitro the core enzyme and the specificity factor are sufficient for correct transcription of mitochondrial template DNA containing a promoter element (Schinkel et al., 1987; Jang and Jaehning, 1991; Xu and Clayton, 1992). Preliminary data from higher eukaryotes indicate that organellar RNA polymerases are generally dependent on accessory factors (Diffley and Stillman, 1991; Fisher et al., 1991; Parisi and Clayton, 1991; Antoshechkin and Bogenhagen, 1995; Bogenhagen, 1996; Larsson et al., 1996; Börner et al., 1999; Bligny et al., 2000).

In this study, detailed heterologous complementation experiments with yeast and plant mitochondrial RNA polymerases were used for the first time to address the question of functional conservation of the core enzymes from different species. The well-established genetic system of yeast (Oliver, 1996; Botstein et al., 1997), its defined mitochondrial transcription apparatus (Grivell, 1995), and the mitochondrial core enzyme from the plant model organism A. thaliana (Hedtke et al., 1997) were combined in these experiments. This approach gives new insights into the function and evolution of mitochondrial core enzymes of today.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strains and Plasmids

A summary of the data is given in Tables 1 and 2.

Yeast strains were grown at 28 or 36°C in glucose or glycerol complete medium (2% glucose or glycerol, 1% peptone, 1% yeast extract, and appropriate amino acids or nucleotides) or minimal medium (2% glucose, 0.67% yeast nitrogen base, phosphate buffer, pH 6.2, and appropriate amino acids or nucleotides). For plates 0.2% agarose was added to the medium.

Escherichia coli strain DH5- (Hanahan, 1983) was used for cloning experiments and amplification of plasmid DNA.

PCR Amplification

Fragments of RPO41 were amplified from the cloned wild-type gene (Lisowsky et al., 1996) by standard protocols (Innis et al., 1990) using standard protocols (Innis et al., 1990), the Taq polymerase kit (TaKaRa), and the primers listed in Table 3. The complete yeast mitochondrial COXII gene was isolated by PCR amplification with the listed primers using purified mitochondrial DNA or whole yeast cells (Sathe et al., 1991) as template sources.

Preparation of Yeast Total DNA and RNA

Total DNA from yeast cells was isolated as described previously, and total RNA was prepared according to the protocol of Schmitt et al. (1990).

Electrophoresis, Southern blots, and Hybridization

Restricted DNA samples or PCR products were electrophoretically separated in 1.0% agarose gels. Southern blots were hybridized with a mitochondrial DNA probe specific for the COXII gene created by PCR and the listed primers. The petite genome E41 served as a specific probe for the mitochondrial 21S rRNA and its gene (Sor and Fukuhara, 1982). Labeling was done by the random priming method (Roche Mannheim, Germany) and [-³²P]dATP. Hybridization was carried out in 60% formamide, 5× SSPE, 5× Denhardt's, 0.1% SDS, and 100 µg/ml herring sperm DNA at 42°C overnight after 6 h of prehybridization.

Isolation of Mitochondria, Protein Gels, and Antibody Studies

Mitochondrial protein extracts were prepared as described previously (Pratje and Michaelis, 1977). The isolated proteins were separated in gradient 4-12% SDS polyacrylamide gels (Novex/Invitrogen, Carlsbad, CA), blotted to nitrocellulose membranes and tested with antibodies specific for Rpo41p (a kind gift of Dr. G.S. Shadel) and yeast CoxIIp. Binding of antibodies was detected by alkaline-phosphatase-conjugated secondary antibodies and chemiluminescence.

Standard Techniques

Plasmid DNA was isolated from E. coli by alkaline lysis and Qiagen Prep Kits. Purification, restriction enzyme digestion, and ligation and analysis of yeast genomic DNA on agarose gels were performed as described (Sambrook et al., 1989). Nucleotide sequences were determined by the biochemical method of Sanger et al. (1977) using T7 polymerase, [-³⁵S]dATP and the appropriate primers or on a sequencing machine AB1373 (Applied Biosystems, Foster City, CA). Intact yeast cells were transformed after lithium acetate treatment according to the procedure of Gietz et al. (1995).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Replacement of Yeast Mitochondrial RNA Polymerase by the Homologous Plant Enzyme

The gene for the full-length mitochondrial core enzyme of A. thaliana (RpoT;1: see Figure 1) was cloned into an expression vector and transformed into a diploid yeast cell that was hetero-allelic for an intact and a deleted copy of the RPO41 gene. The use of such hetero-allelic strains is essential because elimination of the mitochondrial core enzyme from a haploid cell immediately leads to loss of mitochondrial genomes (Greenleaf et al., 1986; Kelly et al., 1986). We also preferred this method to plasmid shuffling because frequent recombination between the gene constructs, especially the wild-type RPO41 gene on the shuffling plasmid, and the chromosomal gene region was observed under these conditions. Therefore, in first experiments we used tetrad dissection to generate haploid yeast mutants for rpo41 that harbored expression plasmids with the respective gene constructs. In addition, tetrad analysis is more selective for Rpo41p-dependent maintenance of mitochondrial genomes (Fangman et al., 1990), because it is known that an alternative mechanism exists that allows Rpo41p-independent replication of specific small mitochondrial DNA fragments (Fangman et al., 1990; MacAlpine et al., 2001). Our results revealed that rpo41 mutant cells were not able to grow on nonfermentable carbon sources (see Figure 2) after transformation with the expression plasmid for the full-length mitochondrial core enzyme from A. thaliana. This indicated that the plant core enzyme was not able to restore normal mitochondrial transcription and the respiratory competence of these cells.

View larger version (35K): [in this window] [in a new window]   Figure 1.   Constructs (C1-C7) for expression of yeast (Rpo41p) and plant (RpoT;1) mitochondrial RNA polymerases in the yeast mutant rpo41. Yeast Rpo41p contains at least nine (I-IX) highly conserved domains (black boxes) that are also present in the plant enzyme from Arabidopsis thaliana (RpoT;1). According to the latest literature some of these domains are further divided into subdomains (Cermakian et al., 1997; Hess and Börner, 1999). Gray bars, plant sequences; white bars, yeast sequences. Restriction sites for exchange of domains and for cloning into yeast expression plasmids are listed. Numbers indicate the first and last amino acid residues of the protein fragments. The first 26 amino acid residues (white box) of Rpo41p are sufficient for efficient import of the protein into yeast mitochondria.

View larger version (54K): [in this window] [in a new window]   Figure 2.   Complementation test for a defective mitochondrial RNA polymerase in yeast. A diploid yeast strain with a normal RPO41 gene and a defective rpo41 gene were used in this study. Test for respiratory competence was done after sporulation and tetrad dissection. Haploid spores with the nuclear rpo41 gene (rpo41::URA3) were identified by growth on glucose minimal medium without uracil (glucose  uracil). Before tetrad dissection the diploid cells were transformed with yeast expression plasmids as indicated. Presence of the plant mitochondrial RNA polymerase or the respective fusion constructs (see Figure 1) is listed in the left margin. Stable replication of the expression plasmid YEp351-ADH and its derivatives was tested on glucose minimal medium without leucine (glucose  leucine). Cotransformation of cells with two different plasmids (YEp351-ADH + p112ANE1) was controlled on glucose minimal medium without leucine and tryptophan (glucose  leucine  tryptophan). Respiratory competence of haploid yeast cells was tested on glycerol complete medium. At least 10 complete tetrads were analyzed for each construct.

Construction of Chimeric RNA Polymerases Harboring Plant and Yeast Domains

The result that the plant enzyme alone is not able to substitute for yeast Rpo41p initiated new experiments with a set of fusion constructs of yeast and plant mitochondrial RNA polymerases. Figure 1 shows the exchange of selected protein domains. The long amino-terminal extension of Rpo41p was always included in the fusion proteins, and only highly conserved domains of the carboxyl-terminal part were exchanged. Furthermore, great care was taken to substitute fragments of identical length at identical positions. All gene constructs were checked by DNA sequencing of the complete reading frames after cloning into yeast expression vectors. These constructs (C1-C5) were transformed into the diploid tester strain with one disrupted copy of the RPO41 gene and investigated for complementation activity of haploid rpo41 mutant cells after tetrad dissection (see Figure 2). It turned out that all of these chimeric genes were unable to restore respiration in the rpo41 yeast mutant.

Investigation of Mitochondrial Genomes in Yeast rpo41 Mutant Cells

The missing respiratory competence of rpo41 mutant cells harboring the chimeric plant genes could either be explained by defective mitochondrial transcription or by a loss of intact mitochondrial genomes. To address this problem, mutant cells were analyzed for their content of functional mitochondrial DNA (⁺). Cells without mitochondrial DNA (°) but with an intact nuclear copy of RPO41 were crossed with the transformed mutant cells. The resulting diploid cells were tested for respiratory competence. These test crosses never generated diploid cells that were able to grow on a nonfermentable carbon source (our unpublished results). The missing respiratory competence of these cells confirmed the absence of intact mitochondrial genomes. To check mutant cells for the presence of mitochondrial petite genomes (), staining with a DNA-specific dye and analysis by fluorescence microscopy were performed. Most of the mutant cells were devoid of any mitochondrial DNA (°). One important exception were mutant cells transformed with gene construct C5. In this construct a carboxyl-terminal fragment of Rpo41p was replaced by the respective plant enzyme domain. All mutant cells containing this fusion construct exhibited a mitochondrial fluorescence typical for genomes. Petite genomes were maintained in nearly all mutants cells even after longer growth and successive inoculations into fresh glucose medium.

C-terminally Truncated Rpo41p Supports Stable Maintenance of Petite Genomes

To test whether the enzyme domain of the plant mitochondrial RNA polymerase at the carboxyl-terminus of construct C5 contributed to the maintenance of deleted mitochondrial genomes, the reading frame of this domain was eliminated from the gene construct. It turned out that truncated yeast Rpo41p comprising amino acid residues 1-974 (C6; Figure 1) was sufficient for maintenance of deleted mitochondrial DNA. To verify that replication of petite genomes was dependent on the expression of truncated mitochondrial Rpo41p, we eliminated the plasmids from the rpo41 mutant cells. This was accomplished by incubation in glucose complete medium and subsequent selection of cells that spontaneously had lost all their plasmids. These yeast cells rapidly lost all mitochondrial genomes immediately after elimination of the gene construct C6 (Figure 3D).

View larger version (174K): [in this window] [in a new window]   Figure 3.   Fluorescence analysis of yeast cells for maintenance of mitochondrial DNA. Yeast cells were stained with the DNA-specific dye DAPI. The nucleus gives the brightest signal in each cell. The presence of mitochondrial genomes is indicated by additional peripheral fluorescence signals. (A) Wild-type, (B) mutant rpo41, (C) mutant rpo41 transformed with the expression plasmid for construct C6, (D) cells from C after elimination of the yeast expression construct C6.

Characterization of the Mitochondrial Petite Genomes in the Mutant Cells

Total DNA was isolated form haploid mutant yeast cell cultures directly after tetrad dissection. Southern blots of restricted DNA were hybridized with a labeled mitochondrial DNA probe specific for the COXII gene region (see Figure 4). This DNA probe was selected, because it contains a small but well-characterized protein-encoding gene and a functional ori/rep sequence element (Foury et al., 1998). The COXII probe identified a mitochondrial DNA fragment of ~20 kb in wild-type cells and the mutant transformed with construct C6, whereas mutant rpo41 is devoid of any mitochondrial DNA. The same result was obtained when a gene probe specific for the mitochondrial large rRNA (21S rRNA) was used. After longer incubation of the petite genome-containing mutant cells in fresh glucose medium instability of the petite genomes was observed but never the complete loss of all mitochondrial DNA fragments (our unpublished results).

View larger version (54K): [in this window] [in a new window]   Figure 4.   Analysis of mitochondrial DNA, RNA, and protein from wild type, the rpo41 mutant and the rpo41 mutant transformed with construct C6 (rpo41 + C6). Total DNA was isolated from yeast cells and aliquots of 10 µg were restricted with BamHI. After separation in a 1% agarose gel Southern blots were prepared and hybridized with gene probes representing the mitochondrial COXII and 21S rRNA genes. These probes identified a higher molecular weight band of ~20 kb in the wild type and in the mutant transformed with C6 but not in the rpo41 mutant. Total RNA extracts from yeast cells were tested with the same mitochondrial gene probes (see MATERIAL AND METHODS). COXII transcripts and 21S rRNA are only found in wild-type cells. For immunological studies aliquots of 20 µg mitochondrial protein were separated by SDS-PAGE. The Western blot was tested with a polyclonal antibody specific for subunit 2 of cytochrome oxidase (CoxIIp). Only wild-type cells expressed detectable mitochondrial CoxIIp.

The presence of the 21S rRNA and COXII gene on the newly formed petite genome prompted us to look for transcripts and proteins. Northern Blots with total RNA extracts were tested with the COXII and 21S rRNA gene probes. Only the wild-type RNA extracts exhibited transcripts for COXII mRNA and 21S rRNA. Crude mitochondrial protein fractions were prepared and aliquots were analyzed in Western blots. Mitochondrial CoxIIp was only present in wild-type protein extracts.

Finally, the petite genomes generated in mutant cells harboring truncated Rpo41p were tested for hypersuppressivity. The phenomenon of petite suppressivness is associated with a special class of deleted mitochondrial genomes that rapidly replace mitochondrial ⁺ DNA from wild-type cells, thereby causing a respiratory defect (Fangman et al., 1989, 1990). In crosses with wild-type ⁺ cells and rpo41 mutant cells harboring the mitochondrial genomes, we determined the percentage of respiratory-deficient diploid cells. More than 99% of the generated diploid cells still were able to grow on a nonfermentable carbon source even after longer incubation and successive replica plating (our unpublished results). This finding argued against the generation of high-suppressive petite genomes in the rpo41 yeast strain transformed with the chimeric gene construct C6.

Tetrad Analysis and Plasmid Shuffling Show Comparable Results

The generation of haploid tester strains by sporulation avoids the problem of generating special petite genomes that are independent of Rpo41p, because this method is more stringent than plasmid shuffling (Fangman et al., 1990). To test if plasmid shuffling leads to different results, the most important constructs C2, C5, and C6 were selected for new plasmid shuffling experiments and compared with the data from tetrad analysis. Wild-type Rpo41p was expressed from the vector p112ANE that contains the tryptophan marker (Riesmeier et al., 1992). This plasmid was combined with the constructs C2, C5, or C6 in a haploid rho⁺ yeast cell with the rpo41 deletion in the nuclear genome. Elimination of the plasmids from the haploid cells was achieved by incubation in glucose complete medium. After elimination of the plasmid with the wild-type RPO41 copy, the mutant cells were tested for respiratory competence and mitochondrial DNA content. It turned out that under these conditions all rpo41 cells were respiratory deficient. The cells containing C2, like a control without any plasmid, lost most of the mitochondrial petite genomes rapidly. To demonstrate the clear differences between Rpo41p-dependent and -independent petite genome maintenance, we designed a new PCR test with whole yeast cells. The maintenance of the complete COXII region in the first tetrad dissection experiments prompted us to use primers for the complete gene in this test. It turned out that only in rpo41 mutant cells containing the constructs C5 or C6 was stable maintenance of larger petite genomes spanning the complete COXII gene region observed (see Figure 5). These experiments verify that the truncated Rpo41p enzyme is responsible for the formation of a specific class of large petite genomes. These newly formed petite genomes are completely dependent on truncated Rpo41p for stable maintenance in haploid rpo41 cells. In these experiments no differences were observed whether using plasmid shuffling or tetrad dissection.

View larger version (25K): [in this window] [in a new window]   Figure 5.   Whole cell PCR amplification of the complete COXII gene for comparison of tetrad dissection and plasmid shuffling experiments. Amplified PCR products were analyzed in a 1% agarose gel. The presence of a mitochondrial genome fragment spanning this region is indicated by the amplification of a 800-base pair DNA fragment. Haploid yeast cells with a nuclear rpo41 were generated either by tetrad dissection or by plasmid shuffling as indicated. Plasmid constructs that were still retained in the cells are listed. Elimination of these plasmids in a second step is indicated (). The approximate generation times for incubation in glucose medium after spore germination or plasmid elimination are listed. An aliquot of the yeast colonies was directly used for whole cell PCR amplification with primers for the complete reading frame of the COXII gene. Controls without any plasmids are: rho° strain (rho°); rho+ wild-type strain (WT).

The Carboxyl-terminal Deletion of Rpo41p Can Be Complemented in trans

The finding that the carboxyl-terminal domain of yeast mitochondrial Rpo41p is essential for stable maintenance of intact mitochondrial genomes prompted us to test whether expression of the missing yeast protein fragment from a second plasmid could reconstitute a functional RNA polymerase inside mitochondria. To direct this protein fragment into yeast mitochondria, the mitochondrial import sequence of Rpo41, consisting of the first 26 amino acid residues of the core enzyme (Wang and Shadel, 1999), was fused to amino acids 903-1351 (C7; Figure 1). Complementation tests with any of the two plasmids alone never restored the respiratory function (our unpublished results). In contrast, coexpression of the two fragments of Rpo41p (C6; C7; Figures 1 and 2) in the mutant rpo41 restored normal respiratory functions at 28°C. The complementation activity of these two protein fragments was not identical to the full-length, wild-type core enzyme because a shift of these cells to higher temperature resulted in a block of respiratory function (see Figure 2). As a final control for complementation in trans, the plasmids from the complemented yeast mutant cells were again eliminated by longer growth in nonselective glucose complete medium. Spontaneous loss of the construct C6 and/or C7 always revealed the original rpo41 mutant phenotype (our unpublished results).

Detection of Mitochondrial RNA Polymerase Fusion Proteins in Yeast Mitochondria

A precondition for functional complementation of mitochondrial transcription in rpo41 mutants is the correct cytosolic expression of the gene constructs on the plasmids and import of the fusion proteins into mitochondria. To test this, crude mitochondrial protein extracts were prepared and separated by SDS-PAGE. After transfer to nitrocellulose, the filter was tested with an antibody specific for Rpo41p (Figure 6). Proteins with the expected molecular weights were detected in the mitochondrial fractions. Mutant cells harboring the constructs C6 and/or C7 for fragments of Rpo41p expressed two distinct proteins of the expected molecular weights. These results confirm a strict correlation between elimination of the plasmids, loss of the respective Rpo41p fragments, and a block in respiratory function. Mutant cells transformed with a gene construct encoding just one fusion protein of yeast and plant enzymes expressed only one higher molecular weight protein that copurified with mitochondria. Examples for fusion constructs C2 and C4 are included in Figure 6.

View larger version (70K): [in this window] [in a new window]   Figure 6.   Detection of RNA polymerases in isolated mitochondria of mutant cells. Protein extracts from isolated mitochondria were separated in a SDS polyacrylamide gel, and the Western blot was probed with a polyclonal antibody specific for yeast Rpo41p. Coexpression of construct C6 + C7 in the yeast mutant rpo41 allowed the detection of two distinct protein fragments with the expected molecular weights. After elimination of construct C6 from these cells the protein fragment with the corresponding molecular weight was no longer present. Examples for expression of a single-subunit fusion protein with domains from yeast and plant mitochondrial RNA polymerases are given for constructs C2 and C4. For detailed description of these constructs see Figure 1.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In eukaryotes, single-subunit RNA polymerases are the hallmark of mitochondrial transcription (Cermakian et al., 1996; Hess and Börner, 1999). Our studies demonstrate the specific adaptations for organellar transcription and give new insights into the function of the C-terminal domains for replication of mitochondrial genomes. For the first time it is shown that a transcriptionally severely impaired Rpo41p fragment still supports replication of petite genomes and that this fragment of Rpo41p can complement in trans, upon coexpression with the missing protein domains. These data can be summarized in a new model for the evolution of single-subunit RNA polymerases by gene fusion events.

First, we investigated functional conservation of the core enzymes form yeast and plants by heterologous complementation experiments. The failure of the full-length plant enzyme to functionally substitute for Rpo41p in vivo could be explained by the high sequence divergence in the amino- and carboxyl-terminal regions of the mitochondrial enzymes from different species (Wang and Shadel, 1999). Therefore, the really surprising result was the finding that none of the chimeric constructs was able to replace Rpo41p. This is in contrast to many successful heterologous complementation experiments in yeast with diverse plant and human enzymes (Minet and Lacroute, 1990; Riesmeier et al., 1992; Lisowsky et al., 1995; Lange et al., 2001). Especially the functional exchanges of components from the multisubunit RNA polymerases of the nucleus have already been documented even for only distantly related species like yeast and humans. (Shpakovski et al., 1995).

Yeast mitochondrial single-subunit RNA polymerase exhibited a high sensitivity against any heterologous domain exchange. The substitution of just one small domain in the highly conserved central part of the protein already resulted in complete loss of respiratory function. According to previous experiments deletion of RPO41 from yeast cells has two effects: 1) block of mitochondrial transcription and 2) loss of intact mitochondrial genomes (Greenleaf et al., 1986). Neither were the exchanged plant enzyme domains able to complement for any of the effects of RPO41 deletion, nor was the C-terminal plant fragment of construct C5 important for the maintenance of the deleted mitochondrial genomes in the rpo41 mutant, because deletion of the plant fragment from the Rpo41p fusion protein revealed a comparable phenotype. Defective mitochondrial respiration in these cells is direct evidence that the plant enzymes and chimeric fusion proteins are not able to restore normal mitochondrial transcription and maintenance of intact mitochondrial genomes.

Maintenance of mitochondrial genomes in the rpo41 mutant was strictly dependent on the presence of truncated Rpo41p, whereas all the other chimeric yeast and plant enzymes did not promote replication of any petite genome. Analysis of the petite genome from haploid mutant cells directly after tetrad dissection identified mitochondrial DNA fragments of ~20 kb that contained the region including the genes for COXII and the 21S rRNA (Foury et al., 1998). It was not possible to identify any mitochondrial transcripts for COXII or the 21S rRNA in the mutant and consequently, no protein for CoxIIp was identified. The failure of truncated Rpo41p to synthesize mitochondrial transcripts reflects the essential function of the missing conserved domains VI-IX that are indispensable constituents of the catalytic core (Osumi-Davis et al., 1992; Bonner et al., 1992, 1994a, 1994b). This indicates that a normal transcription is not essential for replication of mitochondrial genomes. Comparable results were obtained in former experiments with temperature-sensitive yeast Mtf1p. Block of mitochondrial transcription at the restrictive temperature for mutated Mtf1p did not result in any reduction of mitochondrial DNA for several generations (Cliften et al., 1997). Although, at this point of our analysis it cannot be excluded that truncated Rpo41p is still able to synthesize very small amounts of extremely short transcripts for initiation of mitochondrial DNA synthesis. In case that truncated Rpo41p would indeed be completely inactive for RNA synthesis, one can speculate that this protein fragment still supports the replication process either by interactions with protein components of the mitochondrial DNA replication machinery or by changing the conformation of ori/rep sequences as already suggested earlier (Schinkel and Tabak, 1989). First evidence for new protein interactions between the core enzyme and other mitochondrial proteins is the recent identification of a coupling mechanism between mitochondrial transcription and other components of the mitochondrial RNA metabolism by direct interaction of Rpo41p and Nam1p/Mtf2p in yeast mitochondria (Rodeheffer et al., 2001). Future work will have to determine whether also components of the replication machinery interact with Rpo41p, but nevertheless our work demonstrates that even truncated and at least transcriptionally severely impaired Rpo41p supports maintenance of petite genomes. These newly generated petite genomes exhibited instability as already observed for rpo41 mutants (Lorimer et al., 1995). Although, as long as truncated Rpo41p was present, mitochondrial DNA fragments were never completely lost from the mutant cells. The finding that these newly generated petite genomes do not show a hypersuppressive phenotype argues for the enzymatic inactivity of truncated Rpo41p, because latest results from the analysis of hypersuppressive petites demonstrate that this phenomenon depends on a functional Rpo41p and a high density of ori sequences generated by a large number of short repeats form the same ori (MacAlpine et al., 2001). In addition, the work of MacAlpine et al. (2001) identified a hierarchy of mitochondrial promoters that may explain the generation of genomes in our experiments with truncated Rpo41p. C-terminally truncated Rpo41p obviously still supports the function of some of the ori/rep sequences. The failure to synthesize transcripts from or to interact with all of the ori/rep sequences may induce loss of these mitochondrial DNA regions. In our experiments no differences were observed whether using tetrad dissection or plasmid shuffling for the respective experiments.

These data prove that the intact carboxyl-terminal extension of yeast Rpo41p is essential for maintenance and expression of intact mitochondrial genomes and cannot be functionally substituted by the homologous plant domains. How can these negative results for the heterologous complementation experiments with mitochondrial single-subunit RNA polymerases be explained? Several independent lines of investigation point to specific intramolecular protein-protein interactions associated with the different domains of the core enzyme. Two-hybrid and mutation studies with Rpo41p and the specificity factor Mtf1p demonstrated that only the full-length core enzyme facilitated stable protein-protein interactions with Mtf1p and that multiple regions encompassing the entire length of the core enzyme are involved in these interactions (Cliften et al., 1997, 2000). Additional evidence for special intramolecular protein-protein interactions between the domains of Rpo41p is derived from the complementation in trans. In contrast to the high sensitivity of Rpo41p to substitution of any domains by homologous plant enzyme fragments, the physical division of Rpo41p into separated but still functional protein fragments worked surprisingly well (Wang and Shadel, 1999).

Our experiments demonstrate that not only the amino-terminal extension (Wang and Shadel, 1999) but also the carboxyl-terminal domain can complement in trans a respectively truncated Rpo41p. Furthermore, for the first time, highly conserved domains of the supposed catalytic core of Rpo41p are shown to be able to complement in trans. For functional complementation the two separated protein fragments must still support close physical interactions to assemble a transcriptional active RNA polymerase inside mitochondria. Temperature sensitivity of this activity points toward a reduced stability of the protein complex formed by the two separated fragments of Rpo41p.

What could be the basis of such specific interactions between domains of the core enzyme Rpo41p? Our data indicate that complementation in trans by assembly of the two protein fragments into a functional complex inside mitochondria may resemble an early situation during the evolution of these RNA polymerases. We assume that mitochondrial single-subunit RNA polymerases of today are the result of several gene fusion events comparable to the well-documented example of FASI and FASII complexes for fatty acid biosynthesis (McCarthy and Hardie, 1984). In prokaryotes, a multisubunit FASI complex was found that was transformed into the large single-subunit FASII complex of eukaryotes by gene fusion events (McCarthy and Hardie, 1984; Schweizer et al., 1984; Mohamed et al., 1988; Schneider et al., 1997). Comparable events for interacting domains of ancient multisubunit RNA polymerases can be postulated for the generation of single-subunit RNA polymerases of the bacteriophage-type. Adaptation and fine-tuning of these enzymes for mitochondrial transcription would have resulted in very specific intramolecular interactions for each species. According to this model the divergent amino- and carboxyl-terminal extensions from different species would represent different former accessory factors that have been recruited for adaptation to organellar transcription machineries (see Figure 7). Species-specific recruitment of these accessory factors would result in significant differences in organellar transcription regulation. This could also explain why the currently available sequences of mitochondrial specificity factors for the core enzymes exhibit high degrees of divergence (Bogenhagen, 1996; Jan et al., 1999). Already among closely related yeast species, many sequence variations for the specificity factor Mtf1p can be observed (Carrodeguas et al., 1996; Jan et al., 1999). Even more, the mitochondrial transcription apparatus of higher eukaryotes seems to have recruited completely different factors for regulation of transcription initiation (Maliga, 1998; Börner et al., 1999; Hess and Börner, 1999; Bligny et al., 2000).

View larger version (26K): [in this window] [in a new window]   Figure 7.   Model for possible gene fusion events that generated yeast mitochondrial RNA polymerase of today. A hypothetical ancient multisubunit RNA polymerase was the ancestor of the bacteriophage-type enzymes. The bacteriophage enzymes represent a minimal form of the single-subunit RNA polymerase harboring 9 (I-IX) highly conserved domains. Adaptation of this enzyme for mitochondrial transcription required interactions with accessory factors that subsequently were fused with the core enzyme. After transfer of the gene into the nucleus of the eukaryotic cell, the attachment of an import sequence for mitochondria was necessary. Our new model does not represent the exact sequences and order of events but illustrates the possible principles that could have created such fusions.

Our data give new insights into the adaptation of single-subunit RNA polymerases to the very specific conditions of organellar transcription. The importance of the core enzymes for the development of alternative regulatory mechanisms inside mitochondria is stressed by the newly identified coupling mechanism for mitochondrial transcription and RNA processing in yeast (Rodeheffer et al., 2001). The specific interaction of the amino-terminal extension of Rpo41p with Nam1p/Mtf2p is the molecular basis for this phenomenon. Future work will have to determine the influence of gene fusion events and highly specific intramolecular protein interactions on the evolution of diverse functions associated with the core enzymes for mitochondrial and plastid transcription.