A complete sequence of Saccharomyces paradoxus mitochondrial genome that restores the respiration in S. cerevisiae


Correspondence: Pavol Sulo, Department of Biochemistry, Faculty of Natural Sciences, Comenius University, Mlynská Dolina, Bratislava 84215, Slovakia. Tel.: +421 2 6029 6611; fax: +421 2 6029 6452; e-mail: sulo@fns.uniba.sk


We determined the complete sequence of 71 355-bp-long mitochondrial genome from Saccharomyces paradoxus entirely by direct sequencing of purified mitochondrial DNA (mtDNA). This mtDNA possesses the same features as its close relative Saccharomyces cerevisiae – A + T content 85.9%, set of genes coding for the three components of cytochrome oxidase, cytochrome b, three subunits of ATPase, both ribosomal subunits, gene for ribosomal protein, rnpB gene, tRNA package (24) and yeast genetic code. Genes are interrupted by nine group I and group II introns, two of which are in positions unknown in S. cerevisiae, but recognized in Saccharomyces pastorianus. The gene products are related to S. cerevisiae, and the identity of amino acid residues varies from 100% for cox2 to 83% for rps3. The remarkable differences from S. cerevisiae are (1) different gene order (translocation of trnF-trnT1-trnV-cox3-trnfM-rnpb-trnP and transposition of trnW-rns), (2) occurrence of two unusual GI introns, (3) eight active ori elements, and (4) reduced number of GC clusters and divergent intergenic spacers. Despite these facts, the sequenced S. paradoxus mtDNA introduced to S. cerevisiae was able to support the respiratory function to the same extent as the original mtDNAs.


The best known member of the genus Saccharomyces is Saccharomyces cerevisiae, the most widely studied eukaryotic model organism. Owing to its ability to ferment sugars and produce enormous amounts of ethanol this yeast, together with its hybrids, has been used for brewing and baking for millennia and has had a long association with human activity (Pretorius, 2000), leading to its domestication (Fay & Benavides, 2005) Its sister species, Saccharomyces paradoxus, has no domestic uses, and it is considered to be relatively unaffected by human interaction, as it can be found in exudates of oak tree bark and also in samples of oak-associated soil worldwide. Although these two species diverged from a common ancestor approximately 5 million years ago, they are very similar (Kellis et al., 2003). The haploid form of S. paradoxus is composed of 16 linear chromosomes, with a total genome length of approximately 12 million base pairs, and an anticipated 6504 total genes (Scannell et al., 2011). The genome is highly conserved in its coding regions compared with S. cerevisiae, exhibiting 90% identity and 80% in its intergenic regions (Kellis et al., 2003). Interspecies hybrids between S. paradoxus and S. cerevisiae are almost completely sterile and only a few F1 spores are able to germinate. Therefore, these species are increasingly used to test more generalized ecological and evolutionary hypotheses, to understand the processes of speciation (Greig et al., 2002; reviewed in Greig, 2009; Liti et al., 2009).

While more than ten nuclear genomes of S. paradoxus have been sequenced and assembled (Liti et al., 2009), data concerning the mitochondrial genome of S. paradoxus are fragmented and derived from different strains, most especially from the synonymous Saccharomyces douglasii. The size of mitochondrial DNA (mtDNA) has been estimated at 67 kbp with a gene order that differs from S. cerevisiae by a single transversion (Groth et al., 2000). Sequences are known only for cox1 and cob (Tian et al., 1991, 1993), for the genes involved in protein synthesis (Ragnini et al., 1991; Cardazzo et al., 1997; AJ404228; X95973), and also for atp8 and atp9 (Groth et al., 1999).

The aim of this work was to obtain the complete mitochondrial genome sequence of S. paradoxus type strain CBS 432 and to examine its mitonuclear compatibility inside the S. cerevisiae cell.

Materials and methods

mtDNA sequencing

The S. paradoxus strain CBS 432 (NRRL Y-17217T) was obtained from the CBS collection in Utrecht in the Netherlands. This is used as the taxonomic standard, and the superscript T indicates the type strain. Yeast cells were grown at 28 °C in the YPD medium of 2% glucose, 1% yeast extract, and 1% peptone.

The entire mitochondrial genome from S. paradoxus was obtained exclusively by primer-walking of approximately 200 reads, as described in Procházka et al. (2010). The mtDNA was purified mainly by differential centrifugation (Defontaine et al., 1991) or alternatively by gradient centrifugation (Lang & Burger, 2007), and it was directly sequenced using dye terminator–sequencing chemistry (200 ng DNA, 1 pmol primer) on a Genetic Analyzer (Platt et al., 2007; Applied Biosystems, Foster City, CA; ABI310 and ABI3100-Avant). This usually provides good-quality reads at an average of 700 nt long. The initial primers were designed from the mtDNA sequences known for S. paradoxus (syn. douglasii; Tian et al., 1991, 1993; Ragnini et al., 1991; Cardazzo et al., 1997; AJ404228; X95973) or S. cerevisiae (Foury et al., 1998). Then, a set of consecutive primers was designed to ensure at least a 50-nt overlap. The details and pitfalls concerning primer design and mtDNA-sequencing technique will be described elsewhere. The determined mtDNA sequence was deposited in the GenBank under accession number JQ862335.

Annotation and mtDNA analysis

The sequences were assembled by the vector nti advance v. 8.0 software package of Informax, Inc. Searches similar to the already known and annotated mtDNA genome of S. cerevisiae (Foury et al., 1998) were carried out with the use of variations of BLAST. The tRNAs were searched using the program tRNAscan-SE 1.1 (Lowe & Eddy, 1997) and annotated according to S. cerevisiae (Foury et al., 1998). Gene nomenclature followed the rules described in gobase (O'Brien et al., 2006), and introns were identified using the alternative intron prediction tool of RNA weasel (http://megasun.bch.umontreal.ca/RNAweasel/) (Lang et al., 2007) and by comparison with annotated sequences from several strains of S. cerevisiae and S. paradoxus (Tian et al., 1991, 1993; Foury et al., 1998; Wei et al., 2007). Intron and endonuclease nomenclature follows that of Lambowitz & Belfort (1993), and the intergenic open reading frames (ORFs) are numbered according to Foury et al. (1998). The sizes of ribosomal RNAs (rRNA) were inferred from the model elaborated for their equivalents from S. cerevisiae mitochondria (Konings & Gutell, 1995; Fields & Gutell, 1996).

Cybrid construction and characterization

Yeast interspecies (xeno-mitochondrial) cybrids were prepared by protoplast fusion or by kar cross in the strain MCC109ρ0 (MAT alfa, ade2-101, ura3-52, kar1-1, ρ0) or W3031A ρ0 (MAT a, ade2-1, trp1-1, leu2-3, 112, his3-11.15, ura3-11, can1-100, GAL, psi+, ρ0). Cybrid mitochondrial and nuclear genomes were characterized by karyotyping, genetic analysis and restriction digest, as previously described in Špírek et al. (2000) and Sulo et al. (2003). Respiration and growth rates were analyzed in liquid YPGE (1% peptone, 1% yeast extract, 3% glycerol, and 2% ethanol) at 28 °C as described in Špírek et al. (2000), Špírek et al. (2001), Špírek (2002) and in Sulo et al. (2003).

Results and discussion

General features of genes and the genetic code

Saccharomyces paradoxus CBS 432 mtDNA is a circular-mapping 71 355-bp-long mtDNA. Its size is comparable with other Saccharomyces mtDNA already sequenced – S. cerevisiae (82.0–86.2 kbp; Foury et al., 1998; Wei et al., 2007; Akao et al., 2011) and Saccharomyces pastorianus (70.6 kbp; Nakao et al., 2009), and it is only slightly different from the size determined by restriction mapping (67.1 kbp; Groth et al., 2000). The mitochondrial DNA contains the basic set of genes analogous to S. cerevisiae and S. pastorianus, coding for the components of cytochrome oxidase (cox1, cox2, cox3), cytochrome b (cob), subunits of ATPase (atp6, atp8, atp9), both ribosomal RNA subunits (rnl, rns), the rps3 gene for ribosomal protein, rnpB gene for RNA subunit of RNAse P and the tRNA package (24) (Fig. 1, Table 1). As in other related yeasts from the Saccharomyces–Kluyveromyces complex, S. paradoxus also does not encode for NADH dehydrogenase subunit genes. The presence of the trnR2 gene (anticodon ACG) is consistent with the occurence of rare arginine CGN codons, and it appears to be a Saccharomyces-specific trait, as this gene is absent in mtDNAs in the related yeast species Kazachstania servazzii and Naumovia castellii (Kurtzman, 2003; Langkjær et al., 2003). The overall A + T content is 85.9%, which is the highest recognized in yeasts, but similar to the mtDNAs from related yeasts (Foury et al., 1998; Nakao et al., 2009). This percentage is mainly the result of spacious intergenic regions (45 337 nt), where representation of A and T nucleotides reaches 89.7%, and intronic, plus free-standing ORFs; 12 526 nt, with 82.7%. On the other hand, coding regions of exons of protein- and RNA-coding genes at 13 531 nt possess only 76.3% A + T.

Figure 1.

The genetic organization of S. paradoxus mtDNA. Protein-coding genes are marked with thick red arrows, ribosomal RNA and rnpB genes with green arrows, introns with white rectangles and tRNA genes with blue bars. Gene nomenclature follows the rules described in gobase (atp for ATP synthetase subunits; cox for cytochrome oxidase subunits; cob for cytochrome b, rns for small rRNA ribosomal subunit; rnl for large rRNA ribosomal subunit, trn for tRNA coding genes, rps3 for ribosomal protein and rnpB for RNA subunit of RNAse P).

Table 1. Localization of the genes and replication origins in the Saccharomyces paradoxus mtDNAThumbnail image of

Almost all the genes are transcribed from the same strand (except for trnT1, trnW and rns), which is a typical trait of Saccharomyces mitochondrial genomes.

Owing to significant similarity to their S. cerevisiae counterparts, both protein and RNA coding genes can be easily identified (Foury et al., 1998; Seif et al., 2003). Owing to this relatedness, the genetic code differs from the universal code by TGA being read as tryptophan, CTN as threonine and ATA as methionine (Miranda et al., 2006). Protein-coding genes start with ATG, with the two exceptions of ORF5 and free-standing ORF in the cox1-I5β intron, and TAA is used exclusively as a stop codon. Codon usage in the reading frames shows a strong bias toward codons ending with T or A, as previously observed in the mitochondrial genomes of other yeasts (Sekito et al., 1995; Foury et al., 1998). Codon preferences are identical to S. cerevisiae, with the most profound difference being the favoured use of AUG over AUA for methionine and UUU instead UUC for phenylalanine.

The identity of amino acid residues compared with S. cerevisiae varies from 100% for cox2, atp8, atp9 and several cob/cox1 exons to 83% for rps3 (Table 1).

Introns and free standing ORFs

The three cox1, cob, and rnl genes are interrupted by introns, and a significant part of these carries endonuclease-maturase-like ORFs (Table 1, Fig. 2). Although the majority are group I introns, three of them belong to group II, and only one of these (Spcox1-I1) codes for a protein with the maturase-reverse-transcriptase motif. The majority of intervening sequences is inserted in the same sites as occurs in S. cerevisiae (Table 1, Fig. 2). Related introns also contain ORF in phase with the upstream exon, together with one or two LAGLIDADG dodecapeptide motifs characteristic of the homing endonucleases – RNA maturases (for reviews, see Lambowitz & Belfort, 1993; Haugen et al., 2005; Stoddard, 2006). The exception is the aI5β ORF, which is not fused to the upstream exon. Unlike the ω intron reading frame (Foury et al., 1998), maturase and endonuclease functions are not considered as aI5β requires at least five different proteins for processing (Lambowitz & Belfort, 1993; Watts et al., 2011). Three introns here share more than 94% identity with those in S. cerevisiae, with the exception being the unusual intron cox1-I3β (Tian et al., 1993). Although this was not found in S. cerevisiae (Lambowitz & Belfort, 1993; Foury et al., 1998), it occurs in some other Saccharomyces species – S. pastorianus (interspecies hybrid from S. cerevisiae and Saccharomyces bayanus, Nakao et al., 2009) and S. bayanus, S. paradoxus, S. cariocanus, S. kudriavzevii and S. mikatae (Špírek et al., 2001; Špírek, 2002). The cob gene also possesses four intervening introns, and three of these are at the same positions as in S. cerevisiae. Their relatedness also emphasizes a high degree of homology at 78–96% (Table 1). Only the first intron of the gene is located at an unusual position compared with S. cerevisiae but this is inserted at same position described for S. douglasii (Tian et al., 1991) and also for S. pastorianus (Nakao et al., 2009). Its fragmentary ORF contains the GIY-YIG motif discontinuous with the upstream exon. Apparently, the major distinguishing feature between the sister Saccharomyces species' mitochondrial genomes is the presence of the two cox1-I3β and cob-I1α introns, which have not been observed in S. cerevisiae.

Figure 2.

Comparison of intron insertion sites in S. paradoxus mtDNA with Saccharomyces cerevisiae mtDNA within the (a) cox1; (b) cob genes. Red – group I introns; blue – group II introns; filled – with homing endonuclease (known as mobile in S. cerevisiae); checkered – with ORF (not known as mobile in S. cerevisiae); empty triangles – ORF absent. Saccharomyces cerevisiae mtDNA according to Foury et al. (1998).

Figure 3.

Comparison of gene orders among different Saccharomyces and closely related yeast species Kazachstania servazzii (former Saccharomyces sensu lato). Protein-coding genes are marked with thick red arrows, ribosomal RNA genes and rnpB with green arrows, introns with white rectangles, and tRNA genes with blue bars. Gene nomenclature follows the rules described in gobase (atp for ATP synthetase subunits; cox for cytochrome oxidase subunits; cob cytochrome b, rns for small rRNA ribosomal subunit; rnl for large rRNA ribosomal subunit, rps3 for ribosomal protein and rnpB for RNA subunit of RNAse P). The tRNA-coding genes are marked in capital letters. Gene clusters that differ from Saccharomyces cerevisiae gene order are emphasized by colored ellipses (red and yellow transversions, black, green, and blue transpositions). The opposite orientation of trnT1 is marked by the red arrow.

Group I introns containing homing endonucleases are considered to be a class of mobile elements spread at the DNA level into an intronless allele by a process termed ‘homing’. These have undergone horizontal transfer into different species and mainly to the same genes there in a chain-involved cyclical gain and loss (Cho & Palmer, 1999; Goddard & Burt, 1999). Consequently, intronic ORFs often have a much more similar sequence than proteins encoded by the two host genes (Lang, 1984; Hardy & Clark-Walker, 1991). However, the comparison of S. paradoxus mtDNA with that of S. cerevisiae revealed that their exonic sequences are significantly more conserved than the sequences derived from introns (Table 1). Intron homing is catalyzed by the homing endonuclease which mainly contains the LAGLIDADG motifs (Haugen et al., 2005; Stoddard, 2006). However, these introns and homing endonucleases have different evolutionary histories, and they can exist separately (reviewed in Edgell, 2009; Belfort, 2003). Such free-standing ‘endonucleases’, named ORFs have already been described in mtDNAs of S. cerevisiae and S. pastorianus, and their occurrence is strain dependent. From three ORFs, only the ORF1 orthologue has been found in S. paradoxus, with two LAGLIDADG motives and DNA-binding helix-turn-helix domain (Marchler-Bauer et al., 2011). Similar to the situation in different S. cerevisiae strains, ORF2 is truncated, interrupted by GC clusters, and ORF4 is absent.

In all previously identified small ORFs (Foury et al., 1998), the only region with significant homology to ORF5 has been located at the same region as in S. cerevisiae (between ori and trnE, highlighted in Supporting Information, Data S1). Transcription could be initiated from the promoter upstream from ori2, even though the unusual AUA start codon is recognized as being capable of initiating the translation in S. cerevisiae mitochondria in a less efficient manner (Foury et al., 1998).

Intergenic DNA, GC clusters and origins of replication

The putative ori/rep/tra origins of mtDNA replication in S. cerevisiae are approximately 270 bp in length and are characterized by three conserved GC blocks, A, B, and C, separated by AT-rich stretches (Foury et al., 1998; reviewed in Lecrenier & Foury, 2000). Saccharomyces cerevisiae mtDNA contains seven or eight ori elements, dependent on the strain, but only ori 2, 3, and 5 are active as origins for replication. These three contain a transcription initiation site, called r, which is always found upstream from GC cluster C in the active replication (Lecrenier & Foury, 2000). Large A-/T-rich intergenic regions in S. paradoxus are also interrupted by eight origins of replication (Table 1). These possess an 88–94% identical DNA sequence and have been found within the synonymous S. douglasii strain (Cardazzo et al., 1997) and in the related S. pastorianus (Nakao et al., 2009). Although ori2 is interrupted by the M1 GC cluster, all elements should be considered active, owing to the presence of the upstream transcription initiation site ‘r’ (Data S2A).

The major source of polymorphism in intergenic regions in S. cerevisiae consists of 30–80-nt-long GC-rich sequences called GC clusters (Foury et al., 1998). According to the characteristic primary structures, GC clusters (excluding those in ori) are assigned to seven families that exhibit varying degrees of homology (de Zamaroczy & Bernardi, 1986; Weiller et al., 1989). The content of GC clusters is generally significantly lower in S. paradoxus (Data S3), where the M1 class occurs only nine times compared with 43 times in S. cerevisiae. Although the S. paradoxus M1 GC clusters are present exclusively in the noncoding strand, in S. cerevisiae, these are preferentially in the sense strand. These are also preferentially localized upstream from the transcription initiation site in S. paradoxus, thus indicating their hypothetical function as a type of ‘hard stop’ for the previous unit (Data S3). The number of other classes of GC clusters is also significantly reduced (five from M2 class; eight from M3 class and three from V class; Weiller et al., 1989), and these often occur in pairs with the M1 class.

Three colinear S. cerevisiae mtDNAs sequenced so far show a sequence identity of approximately 98% in much of the mitochondrial genome. This holds true despite their reported size variability of 82.0 kbp by Akao et al. (2011), 86.2 kbp by Wei et al. (2007) and 85.8 kbp by Foury et al. (1998). These genomes differ only in the number of GC clusters and according to the presence or absence of segments from intergenic regions. While gene products of S. paradoxus and S. cerevisiae are well conserved, intergenic spacers are considerably divergent. The most conserved, at 74–84%, are 600- to 1300-bp segments across the 270-bp ori sequences. Sequences of upstream and downstream genes or mutated ORFs were less divergent than other spacers. Here, the 1000-bp cox1 and atp6 segment showed 92% identity, and this agrees with the rate established from coding sequences. In general, the similarity between most of the intergenic segments approaches only 70%, and this is significantly lower than the 90–95% found in the S. cerevisiae strains.

Transcription units and adjacent motifs

The highly conserved sequence motif WTATAAGTA should serve as the yeast mitochondrial transcription initiation site (Christianson et al., 1982; Osinga et al., 1984a; Foury et al., 1998; reviewed in Schäfer, 2005). In S. paradoxus, this motif can be found adjacent to genes −7 to −603 from the beginning of RNA or protein sequence (Data S2A). Distal sequences, however, are often duplicated.

The 3′ termini of mitochondrial mRNAs in budding yeast are specified by the dodecamer motif 5′-AAUAA (U/C) AUUCUU-3′ in the 3′ UTR region (Osinga et al., 1984b; Hofmann et al., 1993; reviewed in Schäfer, 2005). Analogous consensus 5′ AAUAA (U/C) AUUCUU-3′ for the 3′ termini of mRNAs was found in the 3′ UTR regions of S. paradoxus 38–539 nucleotides downstream from the termination codon of the protein-coding genes (Data S2B). Putative transcription units were inferred from the consensus sequences (Table 1), which begin at the transcription initiation site and are terminated by consensus of the terminator or by tRNA genes. This implies a mixed tRNA punctuation model, wherein tRNA processing by RNase P can alternatively provide the 3′ end of the primary transcripts (Ojala et al., 1980; Schäfer, 2005).

Gene order and synteny among Saccharomyces yeasts

In many, often unrelated yeasts such as Candida glabrata (Koszul et al., 2003); Brettanomyces custersianus (Procházka et al., 2010); Schizosaccharomyces pombe, and Schizosaccharomyces octosporus (Bullerwell et al., 2003), all genes are encoded on the same DNA strand, and this has been accepted as a stereotypical feature, especially in small size hemiascomycete mtDNAs (Gaillardin et al., 2012; compare the mitochondrial genomes in the databases http://gobase.bcm.umontreal.ca (O'Brien et al., 2006), www.ncbi.nih.gov/genomes/ORGANELLES/organelles.html).

Although the mitochondrial genome of S. cerevisiae with its unusual trnT1 gene located on the complementary strand was regarded as an exception (Foury et al., 1998), this has been observed in other Saccharomyces (Fig. 3) and in the closely related K. servazzii and N. castellii species (Kurtzman, 2003; Langkjær et al., 2003). A further, quite frequent trait involves the majority of genetic information being carried in one strand, with one or a few genes in the complementary strand. Nevertheless, a broader sampling of yeast species proved that this was a biased view, because most of the other yeast mtDNAs were shown to contain several gene blocks on opposite strands (http://gobase.bcm.umontreal.ca (O'Brien et al., 2006), www.ncbi.nih.gov/genomes/ORGANELLES/organelles.html). The MtDNAs from both S. paradoxus and S. pastorianus belong to this group.

Although the regions of synteny are relatively long, both molecules contain besides trnT1 gene, rns-trnW gene block on opposite strand. Saccharomyces cerevisiae gene order differs from S. paradoxus only in one translocation (trnF-trnT1-trnV-cox3-trnfM-rnpb-trnP) and one transversion event, while S. pastorianus differs by two transpositions and one tranposition–transversion event (Fig. 3). It is very interesting, however, that the cluster containing rns-trnW is mobile in both cases.

Mitochondrial genomes can be disassembled into several conserved gene blocks that mimic transcription units (Data S2; Procházka et al., 2010). An assessment of these conserved gene blocks and the putative transcription units clearly show that although the genes in Saccharomyces were reshuffled, the transcriptional units always remained conserved, and this fact is consistent with previous observations inferred from restriction mapping (Cardazzo et al., 1998; Groth et al., 2000).

Yeasts provide a unique opportunity for understanding the general principles of evolutionary mechanisms such as speciation, as seen in experimental hybrid speciation (Greig et al., 2002; reviewed in Greig, 2009; Dujon, 2010 and Chou & Leu, 2010). Interspecies hybrids between various Saccharomyces species, such as S. cerevisiae/S. paradoxus, are capable of establishing fertile lines after repeated cycles of sporulation and self-fertilization, which by definition provides new species. Much of the species' divergence may result from responses to internal genetic conflicts concerning noncollinear chromosomes (Delneri et al., 2003) or ‘antirecombination’ (Greig, 2009). Consequently, hybrid speciations accompanied by whole genome duplication are considered two general principles explaining how yeast species may have arisen, at least in the Saccharomyces/Kluyveromyces complex (Wolfe & Shields, 1997; reviewed in Dujon, 2010 and Liti & Louis, 2005). Functional mitochondria are essential for sporulation, which is the essential step in hybrid speciation. However, their role has been largely overlooked, because although they are bi-parentally inherited in yeast cells, mitochondrial genomes segregate the homoplasmic progeny during zygote outgrowth (for review Dujon, 1981; Ling et al., 2011). Although the fate of mtDNA has not been examined in hybrid speciation lines, it is believed that progeny inherit only intact mitochondrial genome from single yeast species, as observed in natural or in vitro created hybrids (Marinoni et al., 1999; Rainieri et al., 2008; Solieri et al., 2008).

By analyzing mitochondrial genomes from different yeasts, it appears that their genome size, gene order and their noncoding regions are constantly changing during evolution. The reshuffling of gene order in mitochodnrial genomes is a very specific feature that accompanies speciation events in Saccharomyces and perhaps in all hemiascomycetous yeasts. Even two closely related yeast species possess mitochondrial genomes with a different gene order (Solieri, 2010) [http://gobase.bcm.umontreal.ca (O'Brien et al., 2006), www.ncbi.nih.gov/genomes/ORGANELLES/organelles.html]. Thus, the well-conserved genetic function of mitochondrial genomes contrasts with their perplexing diversity in structure and architecture that is apparently species specific (Clark-Walker, 1992; Mueller & Boore, 2005; Solieri, 2010). The underlying mechanisms of gene order rearrangements are still unclear, but they presumably involve transposition duplication followed by the random loss of redundant genes. This most commonly suggested model has been inferred exclusively from genomic comparison (Clark-Walker, 1992; Boore & Brown, 1998). Intergenic region GC clusters, mobile introns and the ability to generate small petite molecules are considered to be triggers in yeasts (Clark-Walker, 1992; Chou & Leu, 2010; Solieri, 2010).

Why does whole genome sequencing not provide sufficient reads to complete the mtDNA sequence?

In recent years, the automated sequencing technique has enabled sequencing of many yeast genomes from genomic libraries. However, sequenced genome data often does not include the mitochondrial genome sequence. While part of the 175 612 total reads supported S. cerevisiae diploid YJM789 mtDNA (Wei et al., 2007) and 82 005 from 118 699 K7 reads supported mtDNAs from sake strain (Akao et al., 2011), 1600 cloned mitochondrial sequences with 5.8-fold coverage did not provide the complete sequence of a different S. cerevisiae mitochondrial genome (Foury et al., 1998).

We attempted to extract S. paradoxus mtDNA from approximately 165 500 reads, which was 7.7 times the genome's coverage (Kellis et al., 2003). All reads were compared with S. cerevisiae mtDNA (NC_001224; Foury et al., 1998) and with S. paradoxus mtDNA (this work) by blastn, but only 64 of these sequences were of mitochondrial origin. Surprisingly, almost two-thirds were sequences of mitochondrial pseudogenes in the nuclear chromosomes. These are the so-called NUMTs, the NUclear sequences of MiTochondrial origin (Leister, 2005; Sacerdot et al., 2008). Only 24 reads can be considered as mitochondrial, mostly supplying only 95–98% identity and representing only approximately 13.5 kbp. Although the mitochondrial DNA in vivo should constitute approximately 15% of the genomic DNA in the diploid S. paradoxus cell, mitochondrial sequences in silico represent only 0.01%, and this is a 1500-fold difference.

Where only study of the nuclear genome is concerned, genomic DNA libraries are constructed from mtDNA deletion mutants (Cliften et al., 2001; Woolfit et al., 2007). Our data suggest that the elimination of mtDNA is not necessary beause mtDNA in genomic libraries is under-represented.

The question is: why is there this discrepancy? The most plausible explanation is the toxicity of cloned mtDNA segments in Escherichia coli. It is recognized that many genomic regions are difficult to capture in conventional circular plasmids. Repetitive and very AT-rich sequences with 75% or more AT are notoriously difficult to clone (Godiska et al., 2010). This instability in cloned fragments is exacerbated by transcription of the cloned DNA, which can interfere with the plasmid replication or with the expression of the selected marker (Godiska et al., 2010).

Experimental genomics, nucleo-mitochondrial interactions and concluding remarks

Mitochondria have a limited genome, and efficient oxidative phosphorylation requires coadaptation of mitochondrial and nuclear genes, despite divergent tempi and modes of evolution. Therefore, mitochondrial biogenesis is dependent on the interplay of several hundred proteins encoded by the nucleus, with eight proteins and several dozen gene products encoded by Saccharomyces mtDNA (Hess et al., 2009; Merz & Westermann, 2009; Lipinski et al., 2010). Consequently, in interspecific hybrids the mitochondrial genome from one partner does not have to communicate equally well with the nuclear genome of the second partner. This is the reason why the progeny are sterile and usually nonviable. This nucleo-mitochondrial communication represents a general mechanism of Dobzhansky-Muller incompatibility, which determines the reproductive isolation during yeast evolution (Lee et al., 2008; Chou et al., 2010; reviewed in Chou & Leu, 2010 and Solieri, 2010). Such a pair of interspecific incompatibility genes was found when S. cerevisiae chromosomes were replaced with chromosomes from S. bayanus (S. bayanus AEP2 allele S. cerevisiae atp9; Lee et al., 2008).

Systematic study of F2 (S. cerevisiae/S. paradoxus) hybrid sterility caused by cytonuclear incompatibility also revealed a couple of mismatched genes, (MRS1 a nuclear gene for intron splicing factor required for the function of mitochondrial cox1 and AIM22 coding for the ligase required for mitochondrial protein lipoylation (Chou et al., 2010). The same mitonuclear incompatibility event associated with speciation was described when mitochondria from S. paradoxus (syn. douglasii) were introduced into S. cerevisiae. These cytoplasmic hybrids (cybrids) did not respire owing to splicing inability of the cox1-I3β intron which was atypical in this new environment (Tian et al., 1993). The splicing factor coded by the S. cerevisiae MRS1 gene proved unable to splice an S. paradoxus-specific intron, in contradiction to its S. paradoxus ortholog (Herbert et al., 1992).

We also studied the communication ability between the nuclear and mitochondrial genomes from different strains of S. paradoxus (syn. douglasii) and S. cerevisiae by the transplacement of mitochondria from one species to mutants lacking mitochondrial DNA (Špírek et al., 2000; Špírek, 2002; Sulo et al., 2003). While S. cerevisiae mtDNA was able to restore respiration in S. paradoxus to a level close to the original (Sulo et al., 2003), there was only partial re-installation of respiration observed in the opposite direction when mtDNA came from two different S. paradoxus strains (Špírek et al., 2000).

To asses the relatedness and variability of genetic information stored in S. paradoxus CBS 432 mtDNA, S. cerevisiae was re-populated with mitochondria containing sequenced mtDNA. Surprisingly, in this case the CBS 432 mitochondria were largely capable of substituting for S. cerevisiae ones. Although the respiration rate was reduced to 50% of the original S. cerevisiae strain, they exhibited comparable growth rates and growth yield on nonfermentable substrates (Špírek et al., 2001; Špírek, 2002). These features indicates the ability of the sequenced mtDNA to substitute for its S. cerevisiae counterparts, although the cox1-I3ß DNA sequences, reason of cytonuclear incompatibility differed only marginally from other S. paradoxus strains (Data S4). In this case splicing factor is coded in mtDNA as can be inferred from petite mapping and the details of this topic will described elsewhere.

Mutual fundamental S. cerevisiae–S. paradoxus mitonuclear compatibility suggests that (1) mtDNA from the CBS 432 strain possesses the entire genetic information required to basic function in S. cerevisiae, to such an extent that it can be considered a S. cerevisiae variant (or mutant); (2) with the exception of ORF5, the small ORFs are very unlikely to have any function, and they are rather a product of genomics; (3) the species specific gene order in Saccharomyces is more likely the result of a speciation event than a phenomenon involved in mtDNA function; (4) severe divergence in the intergenic sequences has no significant effect on respiratory function, thus suggesting elasticity in mtDNA spacers.


Manolis Kellis (Whitehead/MIT Center for Genome Research, USA) is acknowledged for providing raw sequences and Dr. R. Marshall for reviewing the English text. This work was funded by grants from VEGA 1/3242/06, 1/0327/09, and 1/0360/12. The authors wish it to be known that in their opinion, the first author and the corresponding author should be regarded as joint First Authors.