Nature and distribution of large sequence polymorphisms in Saccharomyces cerevisiae

Abstract

To obtain a better understanding of the genome-wide distribution and the nature of large sequence polymorphisms (LSPs) in Saccharomyces cerevisiae, we hybridized genomic DNA of 88 haploid or homozygous diploid S. cerevisiae strains of diverse geographic origins and source substrates onto high-density tiling arrays. Based on loss of hybridization, we identified 384 LSPs larger than 500 bp that were located in 188 non-overlapping regions of the genome. Validation by polymerase chain reaction-amplification and/or DNA sequencing revealed that 39 LSPs were due to deletions, while 74 LSPs involved sequences diverged far enough from the S288c reference genome sequence as to prevent hybridization to the microarray features. The LSP locations were biased towards the subtelomeric regions of chromosomes, where high genetic variation in genes involved in transport or fermentation is thought to facilitate rapid adaptation of S. cerevisiae to new environments. The diverged LSP sequences appear to have different allelic ancestries and were in many cases identified as S. paradoxus introgressions.

Introduction

Large sequence polymorphisms (LSPs), such as those due to deletions or substitutions, can involve from hundreds to millions of nucleotides and form much of the basis of the evolution of genome structure. Although less common than single nucleotide polymorphisms (SNPs), LSPs can account for a large fraction of the genetic differences between species (Britten et al., 2003) and, potentially containing entire genes and regulatory sequences, can have an important impact on the phenotypic variation within and between species (see e.g. Feuk et al., 2006).

Microarray-based methods provide an alternative to DNA sequencing when genomes of large numbers of individuals need to be characterized and compared. The employed microarrays usually carry oligonucleotide features with sequences matched to a fully sequenced reference genome and allow, besides the identification of single-feature polymorphisms (Borevitz et al., 2003), a precise mapping of LSPs when dense enough (e.g. Schacherer et al., 2007; Schacherer et al., 2009).

Previous studies that used microarrays to characterize genome-wide genetic variation in yeast, identified many LSPs and revealed a preference of LSPs to occur in the subtelomeric regions of chromosomes. This subtelomeric localization bias has been attributed to the enrichment in these regions of nonessential genes that confer an adaptive advantage in specific environments and the increased frequency of nonhomologous recombination facilitated by redundant subtelomeric sequences (Winzeler et al., 2003; Schacherer et al., 2007; Schacherer et al., 2009). However, these earlier studies interpreted absence of hybridization to all or most oligonucleotide features targeted at a specific genomic region as a deletion of that region without validation of the nature of the LSP using an independent method.

In the present study, we characterized the genome-wide distribution of LSPs in 88 S. cerevisiae strains of diverse geographical origins and source substrates using high-density tiling arrays. In addition, we used polymerase chain reaction (PCR)-amplification and DNA sequencing of the regions containing LSPs to validate the underlying nature of the structural variants.

Material and methods

Yeast strains

Haploid or homozygous diploid segregants of 88 S. cerevisiae isolates of diverse geographical origins and source substrates were obtained through sporulation of the original isolates, dissection of the produced spore tetrads and germination of the ascospores using standard methods (Sherman, 1991). All S. cerevisiae strains and isolates were kept in glycerol (15% v/v) at −80 °C or on 1% yeast extract, 2% peptone and 2% dextrose (YPD) agar medium at 4 °C for short-term storage. An overview of the 88 S. cerevisiae strains is given in Table S1. All strains have been deposited in, and should be requested from, the Phaff Yeast Culture Collection (http://www.phaffcollection.org/).

DNA extraction and microarray hybridization

Yeast cultures were grown overnight at 30 °C in 50 mL YPD medium and DNA was subsequently extracted using the Qiagen Genomic-tip 100/G (Qiagen) following the manufacturer’s instructions. DNA fragments of 25–50 bp were obtained by incubation of the extracted DNA with 1U of DNaseI (New England Biolabs) for 2 min at 37 °C in 1× DNase I Reaction Buffer (New England Biolabs). After heat inactivation of the DNase I by incubation of the digestion mixture at 95 °C for 20 min, the digested DNA was analyzed on a 2% agarose gel. The DNA fragments were 3′ end-labeled by incubation with 1 nmol of biotin-11-ddATP (Perkin Elmer) and 20 U of terminal deoxynucleotidyl transferase (New England Biolabs) in 1× NEBuffer 4 (New England Biolabs) for 1 h at 37 °C, and the terminal transferase was subsequently heat inactivated at 75 °C for 25 min. Hybridization of the target DNA onto Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays followed standard Affymetrix protocols for hybridization, washing and staining (Gresham et al., 2006), and the arrays were scanned using the Affymetrix scanner at 0.7 μm resolution. Average hybridization intensities were computed for each oligonucleotide feature with the GeneChip® Operating Software (Affymetrix) using the hybridization intensities of the 9 central pixels. Microarray data were deposited at ArrayExpress (http://www.ebi.ac.uk/arrayexpress/) and are available under accession number E-MTAB-627.

Data analysis

Large sequence polymorphisms were detected by interval analysis performed using Affymetrix Tiling Analysis Software (TAS) v1.1 and the individual “.cel” files. After quantile-normalization across microarrays, LSPs were defined as intervals of overlapping oligonucleotide features covering genomic regions of at least 500 bp with maximum hybridization intensities of 6.5 (log2 scale; empirical threshold based on hybridization intensity levels of negative control features). Gaps of maximum 100 bp containing oligonucleotide features with higher hybridization intensities were tolerated within the LSP intervals.

LSP verification

Genomic regions containing LSPs were PCR-amplified in S. cerevisiae strain S288c and a non-S288c allele containing strain (see Table S3) using oligonucleotide primers designed with PRIMER3 software (Rozen & Skaletsky, 2000) and DNA sequences of the flanking regions in S. cerevisiae S288c. PCRs were performed with the Expand Long Template PCR System (Roche Applied Science) following the manufacturer’s protocol: twenty-five microliter reactions were set up with 150 ng template DNA, 300 nM of each oligonucleotide primer (see Table S3), 500 μM of all four dNTPs and 0.4 μL Expand Long Template enzyme mix in 1× PCR buffer 3. The following PCR temperature profile was used: initial denaturation at 94 °C for 2 min; 35 cycles of 94 °C for 10 s, 55 °C for 30 s, 68 °C for 10 min; final extension at 68 °C for 7 min. PCR products were run on 2 % agarose gels, stained with ethidium bromide and visualized with UV transillumination. Upon succesful amplification, PCR products of which the lengths did not confirm genomic deletions were diluted 10-fold and (partially) sequenced with the ABI PRISM® 3730 DNA Analyzer (Applied Biosystems) using the forementioned oligonucleotide primers and the ABI BigDye® Terminator reaction mixture. In addition, the genomic region containg LSP 144 (see Table S3) was partially sequenced to confirm that reduced length of the PCR product corresponded to a true deletion. The obtained nucleotide sequences were compared with fungal genome sequences using the BLAST Network Service (NCBI, National Center for Biotechnology Service; http://www.ncbi.nlm.nih.gov/BLAST) and the CLUSTAL W algorithm (Thompson et al., 1994) as implemented in BIOEDIT v7.0.9.0 (Hall, 1999). All sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. JN021674-JN021788.

Results and discussion

To examine the distribution of large sequence polymorphisms, including deletions and highly diverged regions, in the S. cerevisiae genome, we hybridized genomic DNA of 88 haploid or homozygous diploid S. cerevisiae strains onto separate Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays. These high-density microarrays contain about 2.5 million 25-mer oligonucleotide features, perfectly matched to the S. cerevisiae S288c reference genome sequence and tiled at 5 bp-resolution, and allow precise mapping of SNPs and structural variants relative to the S288c reference strain (Schacherer et al., 2007; Schacherer et al., 2009).

Based on the coordinates of the intervals corresponding to overlapping oligonucleotide features with background-level hybridisation intensities, we identified 384 different LSPs larger than 500 bp in our sample of 88 yeast strains (see Table S2 for the occurrence of the LSPs in the 88 S. cerevisiae strains). These LSPs varied in size between 500 bp and 45.4 kbp (average of 5.6 kbp; see Figure 1a) and covered 188 nonoverlapping regions in the genome with a preference for the subtelomeres (see Figure 1b). The LSP frequencies ranged from 1 to 90 % across the 88 yeast strains (average of 7 %; see Figure 1b), while the number of LSPs per strain varied between 0, as expected, in strain YJM237 (isogenic with S. cerevisiae S288c) and 85 in strain YJM1252 (average of 28; see Figure 1c).

Figure 1.

Figure 1a. Distribution of the sizes of 384 large sequence polymorphisms (>= 500 bp) in the nuclear genome of Saccharomyces cerevisiae, found by hybridization of genomic DNA of 88 S. cerevisiae strains onto separate Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays.

Figure 1b. Distribution of the chromosomal locations of 384 large sequence polymorphisms (>= 500 bp) across the genome of Saccharomyces cerevisiae; The LSPs were identified by hybridization of genomic DNA of 88 S. cerevisiae strains onto separate Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays and their chromosomal coordinates were rescaled to a 0–100 % range.

Figure 1c. Distribution of the frequencies of 384 large sequence polymorphisms (>= 500 bp) in the nuclear genome of Saccharomyces cerevisiae, found by hybridization of genomic DNA of 88 S. cerevisiae strains onto separate Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays.

Figure 1d. Distribution of the number of large sequence polymorphisms (>= 500 bp) per Saccharomyces cerevisiae strain, found by hybridization of genomic DNA of 88 S. cerevisiae strains onto separate Affymetrix GeneChip® S. cerevisiae Tiling 1.0R Arrays.

PCR amplifications for LSP validation were successful for 113 LSPs (29 %) and 39 of these (35 %; see Table S3) were confirmed to be deletions based on the lengths of the obtained PCR products (see e.g. LSP 144 in Figure 2a). The size of the confirmed deletions varied between 500 bp and 14 kbp (average size of 3 kbp) and they occurred with a frequency between 1 and 63 % (average of 7 %). The number of deletion LSPs per strain varied between 0 and 6 (average of 3), and most strains (95 %) contained at least one deletion LSP. Excluding dubious open reading frames, 26 deletion LSPs disrupted coding sequences and 36 ORFs were completely or partially deleted in at least one strain (see Table S4). While 12 ORFs are uncharacterized, none of the 36 ORFs are known to be essential (Cherry et al., 1998; Winzeler et al., 1999). The disruption of VMA1 is a special case, as it is due to the absence of the homing intein VDE observed in 26 % of the yeast strains used in this study (see LSP 81 in Table S3). VDE encodes a site-specific endonuclease with transcription factor activity (Miyake et al., 2003; Hiraishi et al., 2008) that cleaves VMA1 sequences lacking the VDE sequence to introduce homing of the VDE intein (Gimble & Thorner, 1992), and its presence/absence constitutes a known polymorphism in S. cerevisiae (Gimble & Thorner, 1992; Muller & McCusker, 2009). Ten of the 24 disrupted genes with known functions are involved either in transport (SEO1, FUR4, SUL1, VBA2, PCA1, PHO89, VMA1 and MEP2) or in fermentation (MAL33 and AAD10; see Table S4), which reflects the significance of the subtelomeric regions for adaptation of S. cerevisiae to different environmental conditions (Winzeler et al., 2003; Schacherer et al., 2009). Winzeler et al. (2003) reported the subtelomeres of S. cerevisiae chromosomes to be enriched for genes involved in transport, facilitation, fermentation and C-compound metabolism, and suggested that the shuffling of these genes is an important mechanism to allow adaptation of S. cerevisiae to new food sources. Rapid adaptive change is thereby facilitated by the subtelomeric location of these genes, which is rich in redundant sequences such as the X and Y′ elements that promote nonhomologous recombination (Louis et al., 1994). Although adaptation can lead to genetic divergence, none of the deletions (nor any of the other LSPs) were exclusive to a single source substrate.

Figure 2.

Figure 2a. Clustal W alignment of the DNA sequence of the Saccharomyces cerevisiae S288c allele at the locus between nucleotide positions 1053881 and 1065285 on chromosome 15 and the DNA sequence of the allele at the homologous locus in S. cerevisiae YJM470, illustrating a deletion mutation in the S. cerevisiae genome.

Figure 2b. Clustal W alignment of the DNA sequence of the Saccharomyces cerevisiae S288c allele at the locus between nucleotide positions 62968 and 64615 on chromosome 11 and the DNA sequences of the alleles at the homologous loci in S. cerevisiae YJM248 and S. paradoxus NRRL Y-17217, illustrating the introgression of S. paradoxus DNA fragments into the genome of S. cerevisiae.

For 74 of the 113 succesfully amplified LSP regions (66 %), PCR amplification did not indicate the LSPs to be deletions. Instead, direct sequencing of the PCR products revealed divergence from the S. cerevisiae S288c reference sequence, in the form of a high local frequency of SNPs and small insertions/deletions, which prevented hybridization to the microarray oligonucleotide probes. For 47 of the 74 diverged LSPs, a comparison with all available Saccharomyces genome sequences showed that the highest sequence identities (> 95 %) were with S. paradoxus, which is considerably higher than the average sequence identity between S. cerevisiae and S. paradoxus of about 85 % (Kellis et al., 2003) and which indicates these LSPs to be introgressed S. paradoxus DNA fragments (see e.g. LSP 107 in Figure 2b). These suspected introgressed S. paradoxus fragments/LSPs varied in length between 500 bp and 12 kbp (average of 3 kbp), and occurred with a frequency between 1 and 6 % (average of 2 %; see Table S3). Twenty-three of the 88 yeast strains (26 %) contained at least 1 introgressed S. paradoxus fragment/LSP, and relatively high numbers of such introgressions were identified in strains YJM247 (14), YJM248 (13), YJM1078 (9) and YJM1252 (20). Introgression of S. paradoxus DNA fragments into the S. cerevisiae genome has been described before (Liti et al., 2006; Wei et al., 2007; Muller & McCusker, 2009; Zhang et al., 2010) and has been proposed to be the result of homologous recombination after a hybridization event between the two species (Muller & McCusker, 2009; Zhang et al., 2010). The relatively large number of S. paradoxus sequences dispersed throughout the genomes of strains YJM247, YJM248, YJM1078 and YJM1252 support this hypothesis, as horizontal gene transfer is rare in yeast and large DNA fragments are unlikely to be transferred directly (Dujon, 2005). While many of the strains containing S. paradoxus DNA fragments are of clinical origin, an environment in which S. paradoxus is not commonly found, several strains originated from substrates where both species might coexist and thus form interspecies hybrids (e.g. alpechin, soil and cherry tree; Barnett et al., 2000; Sniegowski et al., 2002; Zhang et al., 2010).

The influence of duplicated sequences on chromosomal stability and the occurrence of LSPs is illustrated by the involvement of sequences similar to Ty elements and to members of the DUP240 multigene family in the flanking regions of several LSPs. The flanking regions of 5 of the 74 diverged LSPs (LSPs # 190, 194, 263, 286 and 306; data not shown) contained sequences similar to Ty elements (transposon yeast). Ty elements are retrotransposons that reintegrate in the genome via RNA intermediates and self-encoded reverse transcriptase and integrase, and are known to induce deletions and other chromosomal rearrangements by allelic or ectopic recombination (Roeder et al., 1984; Rachidi et al., 1999). In the case of 3 LSPs (data not shown), the flanking regions contained sequences similar to the tandem I (LSP # 18) and tandem VII (LSPs # 164 and 166) repeats of the DUP240 multigene family. The DUP240 gene family consists of 10 genes in S. cerevisiae S288c and most ORFs encode membrane-associated proteins of unknown function (Poirey et al., 2002). Just as Ty elements, these genes have been shown to be targets for intra- and interchromosomal recombinations and are involved in chromosomal rearrangements that cause the creation and disappearance of DUP240 paralogs (Leh-Louis et al., 2004).

Forty-three LSPs (12 %; see Table S3) involved the terminal chromosome segments and could therefore not be confirmed by PCR amplification and subsequent DNA sequencing. However, it is noteworthy that 3 strains (YJM1342, YJM1399 and YJM1418) each contained 18 LSPs (average size of about 6 kbp) at one or both ends of most chromosomes while the remaining 85 strains contained at most two of these terminal LSPs. Although these LSPs may consitute diverged sequences that resulted from interchromosomal recombination and that caused loss of hybridization, they may alternativelyhave been deleted by incomplete replication or degradation of the chromosome ends after telomere loss. In S. cerevisiae, as in most eukaryotes, the telomeres consist of simple, repeated DNA that ensures complete replication of the chromosome ends through mediation of the telomerase enzyme (Louis, 1995). Loss of the telomeres, e.g. through inhibition of the telomerase pathway (Zhou et al., 2000), leads to chromosomal instability due to incomplete replication and degradation (Louis, 1995). While absence of telomeres often leads to chromosome loss, chromosome ends can be repaired and stabilized by homologous recombination, nonhomologous end-joining and de novo telomere formation (Mangahas et al., 2001).

To conclude, in our study we used tiling arrays based on the S. cerevisiae S288c reference genome sequence to characterize the LSPs that occur in the nuclear genomes of 88 haploid or homozygous diploid S. cerevisiae strains. While the hybridization profiles suggested many of the identified LSPs to involve deletions, PCR amplification and DNA sequencing revealed that a significant number actually involves alleles with nucleotide sequences that have diverged enough from the reference genome sequence to prevent hybridization to the microarray oligonucleotide probes. While these diverged alleles often indicate different intraspecific ancestries, several were identified as introgressed from S. paradoxus through homologous recombination which indicates likely co-occurrence of S. cerevisiae and S. paradoxus as well as the formation of viable interspecies hybrids. Many of the LSPs are found in the subtelomeric chromosomal regions, where they are less likely to disrupt essential genes. Deletion LSPs thereby often affected genes involved in transport or fermentation, generating genetic variation that allows rapid adaptation of S. cerevisiae to new environments.