A set of genetically diverged Saccharomyces cerevisiae strains with markerless deletions of multiple auxotrophic genes


  • Hélène Louvel,

    1. UPMC, UMR7238, Génomique des Microorganismes, Paris, France
    2. CNRS, UMR7238, Génomique des Microorganismes, Paris, France
    Search for more papers by this author
  • Alexandre Gillet-Markowska,

    1. UPMC, UMR7238, Génomique des Microorganismes, Paris, France
    2. CNRS, UMR7238, Génomique des Microorganismes, Paris, France
    Search for more papers by this author
  • Gianni Liti,

    1. Institute for Research on Cancer and Ageing of Nice (IRCAN), CNRS UMR 7284, INSERM U1081, University of Nice Sophia-Antipolis, France
    Search for more papers by this author
  • Gilles Fischer

    Corresponding author
    1. UPMC, UMR7238, Génomique des Microorganismes, Paris, France
    2. CNRS, UMR7238, Génomique des Microorganismes, Paris, France
    • Correspondence to: Gilles Fischer, Université Paris 06, CNRS UMR7238, Unité de Génomique des Microorganismes, 15 Rue de l'École de Médecine, 75006 Paris, France. E-mail: gilles.fischer@upmc.fr

    Search for more papers by this author


Genome analysis of over 70 Saccharomyces strains revealed the existence of five groups of genetically diverged S. cerevisiae wild-type isolates, which feature distinct genetic backgrounds and reflect the natural diversity existing among the species. The strains originated from different geographical and ecological niches (Malaysian, West African, North American, Wine/European and Sake) and represent clean, non-mosaic lineages of S. cerevisiae, meaning that their genomes differ essentially by monomorphic and private SNPs. In this study, one representative strain for each of the five S. cerevisiae clean lineages was selected and mutated for several auxotroph genes by clean markerless deletions, so that all dominant markers remained available for further genetic manipulations. A set of 50 strains was assembled, including eight haploid and two diploid strains for each lineage. These strains carry different combinations of leu2∆0, lys2∆0, met15∆0, ura3∆0 and/or ura3∆::KanMX-barcoded deletions with marker configurations resembling that of the BY series, which will allow large-scale crossing with existing deletion collections. This new set of genetically tractable strains provides a powerful tool kit to explore the impact of natural variation on complex biological processes. Copyright © 2013 John Wiley & Sons, Ltd.


Yeast, such as S. cerevisiae, has been a powerful tool in eukaryotic genetics, from the description of the information-carrying genome to the association of gene mutation with human disorders. A huge amount of the genetic information that was gained through the study of S. cerevisiae has resulted, more or less directly, from the development of large-scale mutant libraries (Giaever et al., 2002; Mnaimneh et al., 2004; Winzeler et al., 1999; Yu et al., 2006), which initially relied on the design of a set of laboratory strains carrying multiple markerless deletions of auxotrophic genes (Brachmann et al., 1998). These strains have allowed decisive progress to be made in our understanding of many complex biological processes, including cell metabolism, the genetics of mitochondria, meiosis, replication, DNA repair, environmental adaptation and evolution (Botstein and Fink, 2011). However, the extensively studied laboratory strains are not representative of the genetic and phenotypic diversity of the species. Remarkably, extensive cell proliferation phenotyping under many environmental conditions has indicated extreme phenotypes for the laboratory strain S288c when compared to dozens of other S. cerevisiae strains (Warringer et al., 2011). An additional striking example of the S288c peculiarity is the extremely high rate of petite formation, which can affect multiple phenotypes (Dimitrov et al., 2009). Furthermore, recent studies have shown that the same mutation in closely related strains can lead to drastic phenotypic differences as a consequence of background-specific genetic interactions (Chin et al., 2012; Dowell et al., 2010). It is therefore essential to look at diverse backgrounds in order to pursue a comprehensive investigation of genetic polymorphism addressing, especially, how genetic combinations and contexts impact phenotypes. Furthermore, the possibility of exploiting this natural genetics variation is of extreme importance, given the emergence of S. cerevisiae as a model to study natural variation, ecology and population-level genetics (Cromie et al., in press; Hyma and Fay, 2013; Stefanini et al., 2012).

Genome sequencing of over 70 Saccharomyces strains and the study of the phylogenetic relationships within the S. cerevisiae population has revealed five major groups of wild-type isolates collected from diverse sources and referred to as the Malaysian, West African, North American, Wine/European and Sake groups (Liti et al., 2009). These groups were considered to be distinct 'clean' lineages characterized by monomorphic and private SNPs and displayed similar phylogenetic relationships across their entire genome, which contributes to forming consistent groups. The population genomics survey also revealed other strains with high sequence similarities despite distinct worldwide distribution and ecological situations (Liti et al., 2009). Indeed, the domestication of S. cerevisiae strains, e.g. selection and manipulation for human purposes, may have triggered genetic exchanges through strain crosses. Such populations with biased genomic variability, also named 'mosaic' strains, might therefore provide only a limited understanding of the relationship between the genetic background and the function, stability, dynamics and evolution of genomes. In contrast, the five groups feature distinct genetic backgrounds as a reflection of the natural diversity existing among S. cerevisiae species. They represent ideal tools regarding a wide variety of genetic experiments that aim to connect phenotypes to genotypes. This includes the generation of mapping populations for quantitative trait loci analysis (Cubillos et al., in press) and understanding trait divergence underlain by differences in expression levels (Chang et al., 2013; Lee et al., in press) and non-genetic determinants (Ziv et al., 2013).

Representative isolates of each of the five S. cerevisiae clean lineages were previously genetically engineered to make strains readily usable for laboratory purposes (Cubillos et al., 2009). Stable tractable derivatives were thereby developed by disrupting one HO allele with the hygromycin resistance cassette and replacing URA3 with the KanMX cassette containing a 6 bp unique 'barcode'. Furthermore, these strains were characterized at both the genomics and the phenomics level (Liti et al., 2009; Warringer et al., 2011). However, the lack of multiple genetic markers in these new strains has limited their use. In an effort to address this issue and to extend the collection of strains that would be representative of the diversity of the species, we inactivated several of the auxotrophic marker genes commonly used in yeast genetics in the URA3- and HO-mutated derivative strains. First, LEU2, LYS2 and MET15 markerless deletions (deletions of entire ORFs without leaving any sequence from positively selectable marker at the corresponding loci) were made by the pop-in/pop-out method (Brachmann et al., 1998). Then, the KanMX cassette that was originally introduced in URA3 for deletion (Cubillos et al., 2009) was removed by the same technique to allow the recycling of this dominant marker for further constructs. Versions of the LEU2, LYS2 and MET15 auxotroph strains carrying either URA3 replaced by the KanMX cassette, along with a 6 bp 'barcode' (ura3∆::KanMX-Barcode), or URA3 deleted without replacement by an additional selectable marker (ura3∆0), were generated. As a result, a collection of 50 strains of S. cerevisiae was assembled, including eight haploid and two diploid strains for each lineage, with four auxotroph markers (LYS2, LEU2, MET15 and URA3) that can be used for further genetic manipulation. This set of strains will undoubtedly be very useful to explore the impact of natural diversity on phenotypic traits and to understand the genetic mechanisms underlying complex traits.

Materials and methods

Strains and media

The original wild-type yeast strains used in this study were previously reported in Liti et al. (2009) as part of the Saccharomyces Genome Resequencing Project (SGRP). An additional genomic resource was recently released providing de novo genome assemblies (http://www.moseslab.csb.utoronto.ca/sgrp/download.html). The five genetic backgrounds selected here are representatives of the major diverged lineages described in the SGRP, which were designated Malaysian, West African, North American, Wine/European and Sake. Stable haploid (MATa or MATα, hoΔ::HphMX, ura3Δ::KanMX) and diploid (MATa/MATαHO/hoΔ::HphMX, ura3Δ::KanMX/ura3Δ::KanMX) derivatives for one representative isolate of each of the five major lineages were subsequently generated and described (Cubillos et al., 2009). In this study, these HO– and URA3-inactivated derivatives were referred to as 'parental strains' (Table 1).

Table 1. Parental S. cerevisiae clean lineage strains previously described (Cubillos et al.., 2009) and used a starting point for further genetic manipulations
BackgroundOriginal strainGenotypeNCYC Nos
  1. NCYC numbers for haploid strains are indicated between square brackets.

Wine/EuropeanDVBPG6765(Mata/Matα, HO/hoΔ::HYG, ura3Δ::KanMX/ura3Δ::KanMX)3570 [3597: (MATa), 3622: (MATα)]
West AfricanDVBPG6044(Mata/Matα, HO/hoΔ::HYG, ura3Δ::KanMX/ura3Δ::KanMX)3574 [3600: (MATa), 3625: (MATα)]
North AmericanYPS128(Mata/Matα, HO/hoΔ::HYG, ura3Δ::KanMX/ura3Δ::KanMX)3581 [3607: (MATa), 3632: (MATα)]
SakeY12(Mata/Matα, HO/hoΔ::HYG, ura3Δ::KanMX/ura3Δ::KanMX)3579 [3605: (MATa), 3630: (MATα)]
MalaysianUWOPS 03–461.4(Mata/Matα, HO/hoΔ::HYG, ura3Δ::KanMX/ura3Δ::KanMX)3576 [3602: (MATa), 3627: (MATα)]

Routinely, yeast cells were grown on yeast extract, peptone and glucose medium (YPD) (Sherman et al., 1986). For auxotroph selections, cells were grown on complete synthetic medium (CSM) depleted for the appropriate amino acid (lysine, methionine, leucine) or uracil (Sherman et al., 1986). When necessary, YPD was supplemented with hygromycin B 200 µg/ml or geneticin G418 200 µg/ml and CSM was supplemented with 5-fluoro-uracil (5-FoA) 1 g/l. In order to test for petite phenotypes, the 50 auxotroph strains generated in this work were grown on glycerol-containing rich medium.

Transformations of S. cerevisiae strains

Transformation of the yeast cells with DNA templates was performed using the lithium acetate method (Gietz and Schiestl, 2007). Briefly, cells were grown in YPD to exponential phase, washed in water, then lithium acetate 0.1 m, and incubated with 1–2 µg DNA (previously linearized pAD plasmid or PCR product) in the presence of carrier DNA (single-stranded salmon sperm DNA) and polyethylene glycol (PEG). The cells were then heat-shocked at 42 °C for 20 min and plated on selective media.

Sporulation and haploid selection

Diploid cells were sporulated for 2–5 days at 30 °C on 2% potassium acetate agar plates. The cells were incubated for 15 min at 37 °C in a 5 mg/ml zymolyase solution. Spores were dissected using a MSM400 Singer (UK) dissection microscope on YPD agar plates and then incubated for 2 days at 30 °C. The colonies were then replica-plated onto appropriate selective media, such as CSM agar plates lacking lysine, leucine, methionine or uracil, in the case of selection for auxotroph mutant strains, or YPD agar plates supplemented with G418 for ura3∆::KanMX strains. Spores from four-viable-spore tetrads with correct marker segregations were selected. Drug resistances and auxotrophies were obtained according to the expected 2:2 segregation. Phenotypes of the spores were checked by restreaking the strains on selective media and amplifying the deleted auxotroph gene loci by PCR.

Crosses between haploid strains and generation of diploids

The mating types of haploid strains were determined by crossing with MATa and MATα tester strains deleted for the LYS5 gene (lys5∆0). Diploids were selected on minimal medium not supplemented with amino acids or uracil. Growth following crossing with MATa, lys5∆0 indicated that the mating-type of the tested haploid strain was MATα and growth with the MATα tester strain characterized a MATa haploid strain.

To generate diploid strains, two haploid strains with opposite mating types were mixed on YPD agar plates, allowed to grow overnight at 30 °C and restreaked on YPD agar medium to obtain single colonies. In the case of balanced auxotrophies (selection of double- and triple-heterozygous mutant diploids), cells mixed on YPD were replica-plated onto minimal medium lacking all amino acids and supplied with uracil. Diploid colonies were further identified by testing sporulation abilities.

Pop-in/pop-out gene deletions in S. cerevisiae ‘clean’ lineages

UWOPS 03–461.4, DVBPG6044, YPS128, DVBP G6765 and Y12 URA3-deleted diploid strains (ura3∆::KanMX/ura3∆::KanMX) (Table 1) were transformed independently with specific S. cerevisiae marker deletion plasmids (ATCC-LGC) for the complete markerless deletion of auxotroph genes based on the pop-in/pop-out deletion method (Brachmann et al., 1998). The pAD1, pAD2 and pAD4 plasmids, respectively, carry homologous regions flanking LEU2, LYS2 or MET15 and specifically target these auxotroph genes for deletion using URA3 as a selection marker (pAD plasmids carry the URA3 cassette and confer uracil prototrophy). Plasmid integrants were selected on uracil-depleted CSM. The candidates were confirmed by restreaking on uracil-depleted CSM and integration of the plasmid was checked by PCR at one end of the integration site. The positive clones were heterozygous LEU2/leu2::pAD1 or LYS2/lys2::pAD2 or MET15/met15::pAD4.

Complete deletions of the targeted auxotroph genes require excision of the inserted pAD plasmids. Diploid transformed colonies were sporulated and dissected to identify four-viable-spore tetrads. Spores were selected for both hygromycin B resistance and uracil prototrophy and grown overnight in YPD liquid medium to be plated onto 5-FoA-containing CSM. After 2 days of growth at 30 °C, 5-FoA-resistant clones, i.e. uracil auxotrophs, were replica-plated onto leucine, lysine or methionine-depleted CSM, with regard to their respective mutation. Auxotrophies were checked on the appropriate corresponding selective media as well as on uracil-depleted CSM. The complete deletion of the targeted gene (and loss of the URA3-carrying plasmid) was further confirmed by PCR amplification of the targeted locus. According to this protocol, leu2∆0, lys2∆0 or met15∆0 single mutants of mating-types MATa and MATα were generated in each of the five S. cerevisiae lineages. Triple-mutant strains were obtained by successive rounds of crosses between the single mutant strains and sporulation experiments for the five lineages.

Deletion of URA3 and removal of the KanMX cassette

The URA3 wild-type gene (1800 bp, containing the ORF and 500 bp both upstream and downstream) was amplified from the genome of the uracil prototroph S. cerevisiae tester strain (lys5∆0). This DNA amplicon was used to transform the LEU2-, LYS2- and MET15-deleted mutant strains (haploids MATa, ho∆::HphMX, ura3∆::KanMX, leu2∆0, lys2∆0, met15∆0) in order to restore a wild-type copy of URA3. Transformants were selected on uracil-depleted CSM plates. Their uracil prototrophy and G418 sensitivity phenotypes were retested and PCR amplification verified the restoration of the functional URA3 gene.

The pJL164 plasmid (ATCC-LGC) was used as a template to amplify a deleted version of the URA3 gene. This piece of DNA includes both upstream and downstream URA3 ORF flanking regions linked together (without any URA3 ORF sequence or additional marker). The PCR amplicon was used to transform each uracil prototroph mutant (haploids MATa, ho∆::HphMX, leu2∆0, lys2∆0, met15∆0) in order to delete URA3. Transformants were directly plated on YPD after transformation for overnight growth at 30 °C and replica-plated the next day onto 5-FoA plates. Cells were grown for 1–4 days at 30 °C, depending on the strain. Their uracil auxotrophy phenotype was checked on URA drop-out plates and the deletion of URA3 was checked by PCR.

Backcross with parental strains

Mutants carrying all four auxotrophies (MATa, ho∆::HphMX, leu2∆0, lys2∆0, met15∆0, ura3∆0) were backcrossed twice with their respective parental strain (MATα, ho∆::HphMX, ura3∆::KanMX). The first round of backcrossing led to the selection of spores carrying all four auxotrophies, which were further backcrossed once with the parental strains. After the second backcrossing, spores were selected according to the genotypes reported in Table 3. Selected haploid auxotroph strains were subsequently crossed to each other to generate diploid strains with balanced gene markers for the LYS2 and MET15 alleles (Table 3). All strains are available from the National Culture Yeast Collection (http://www.ncyc.co.uk/) and Accession Nos are listed in Table 3.

Evaluation of cell fitness

Growth assays were performed using Tecan Sunrise 96-well microplate readers and data analysis with Magellan software. First, cells were grown to stationary phase at 30 °C in liquid YPD [optical density (OD) = ~20]. Cultures were then diluted and used to inoculate 100 µl fresh YPD or CSM liquid medium at an estimated initial OD = 0.02. Cultures were grown at 30 °C with constant shaking and ODs were measured every 30 min. Sterile medium was used as control. The parental strains (both MATa and MATαho∆::HphMX, ura3∆::KanMX haploids and HO/ho∆::HphMX, ura3∆::KanMX/ura3∆::KanMX diploid) and the S. cerevisiae BY4741 strain were grown along the 50 auxotroph strains in the same conditions. Growth curves were measured in duplicate. Generation times were estimated using a dedicated program (Courbe de Croissance v. 1.4, courtesy of J. Schacherer), which automatically plots (in semi-log scale) the growth curves from the TECAN absorbance data and fits a line to the exponential phase (this fit was manually checked and adapted when necessary). Generation times are automatically calculated from the slope of this line (i.e. the growth rate) as follows: doubling time = ln2/growth rate.

For the Malaysian strains, dry weight experiments were performed in parallel in 10 ml liquid cultures (YPD or CSM). For each strain, four independent cultures were inoculated with either YPD or CSM with the same cell density [100 µl from a small colony (~106 cells) resuspended in 1 ml sterile water]. Each of the four cultures was grown under agitation (160 rpm) at 30 °C up to a specific time (four time points from 800 min to 2400 min for the last time point). At the four time points, cells were filtered on 0.22 µm cellulose filters (GSWP04700, Millipore), which were then dried at 65 °C for 2 h and weighed on a scale. Each experiment was performed in duplicate.

Results and discussion

Four auxotrophic marker alleles, URA3, LYS2, LEU2 and MET15, commonly used in yeast genetics and genomics, were selected for complete markerless deletion. The URA3, LYS2, LEU2 and MET15 genes encode essential enzymes of the uracil, lysine, leucine and methionine biosynthetic pathways, respectively. The four mutations were obtained by pop-in/pop-out (Brachmann et al., 1998), using the pAD plasmid series, leading to complete deletion of the targeted protein coding sequences. Precisely, deletion of LYS2 included the LYS2 ORF, 304 bp upstream of LYS2 ATG and 24 bp downstream of its stop codon. Deletion of LEU2 targeted 6523 bp upstream of ATG and 63 bp downstream of the stop codon; this deletion removed a Ty2 retrotransposon from the DVBPG6765, Y12 and YPS128 strains, while this element is probably absent from the UWOPS 03–461.4 and DVBPG6044 strains, as suggested by SRGP pair-end reads spanning the insertion region (Liti et al., 2009). Deletion of MET15 removed MET15 ORF, 260 bp upstream of its ATG and 811 bp downstream of the MET15 stop codon. URA3 was either replaced from ATG to the stop codon by the KanMX cassette (Cubillos et al., 2009) or deleted from 222 bp upstream of its ATG to 75 bp downstream of its stop codon. The work flow of the strain construction is illustrated in Fig. 1.

Figure 1.

Genealogy of the auxotrophic strains. The successive steps followed in this study are shown. (A) Three auxotrophic genes were independently inactivated in each S. cerevisiae lineage without inserting selectable markers (markerless). (B) They were used to make a triple markerless auxotroph mutant strain. (C) A markerless mutated copy of URA3 (ura3∆0) was made by removing the KanMX-barcoded cassette, which was inserted in the URA3 gene in the parental strains. (D) Backcrossing with the parental strains was realized twice. (E) A set of distinct auxotrophic mutant strains (haploids and diploids) was assembled as indicated

Markerless auxotroph gene deletions in the S. cerevisiae 'clean' lineages

One representative isolate of each of the Wine/European, West African, North American, Sake and Malaysian lineages (ura3∆::KanMX/ura3∆::KanMX) (Table 1), was mutated for LYS2, LEU2 and MET15, independently, using the pop-in/pop-out gene deletion method (Fig. 1A). This two-step process includes the integration of a linearized plasmid (pAD) into its specific chromosomal location by homologous recombination (pop-in). Overall, transformations with the integrative plasmids showed good efficiencies in all strains and for all targeted loci (Table 2), averaging 100–1000 transformants/µg pAD1 DNA, 400–1500 transformants for pAD2 and 50–1500 transformants for pAD4. The Malaysian-derived strain, however, yielded 2–30-fold fewer transformants than the other strains. This could possibly be due to the high level of aggregation of this strain, which could be detrimental to transformation efficiency (see supporting information, Table S1).

Table 2. Semi-quantitative data obtained during the genetic manipulation of the strains
Genetic mechanismCarrierTargeted geneWine/EuropeanWest AfricanNorth AmericanSakeMalaysian
  1. a

    Number of transformants/µg DNA.

  2. b

    Percentage of auxotrophic cells.

Plasmid pop-inapAD1LEU210004002001000100
Plasmid pop-out (%)bpAD1LEU2> 305< 1< 11.5–4
 pAD4MET15710< 1.5< 11–3
Restoration of wild-type URA3aPCR productURA3385481896
Deletion ura3Δ0aPCR productURA366261

Afterwards, the strains that underwent plasmid excision (pop-out), leading to the deletion of the auxotroph gene, were selected, so that single mutants for either LEU2, LYS2 or MET15 were generated in each of the five lineages (Fig. 1A). The observed number of auxotrophic colonies obtained at this step varied greatly, depending on the genetic background and the targeted locus (Table 2). In our hands, the deletion of LYS2 occurred more frequently than the other two deletions in the five strains (1–30%). LEU2 and MET15 deletions highly fluctuated between strains (< 1% to >30% and < 1% to 10%, respectively). However, this result does not seem to correlate with the size of the homology shared between the chromosome and the plasmid, since all homologous regions are similar (~1 kbp). The leucine auxotrophy phenotype was more easily obtained in the Wine/European and North American strains than in the three other strains, whereas mutants for MET15 were more easily identified in the North American, Sake and Malaysian backgrounds. In the other cases, the efficiencies to obtain leucine and methionine auxotrophs were lower.

Furthermore, the strains did show dissimilar sporulation efficiencies on potassium acetate-containing medium (see supporting information, Table S1). NaCl- and malt-based media were also tested, but did not improve sporulation efficiencies. The North American and Malaysian strains had higher rates of sporulation, which were estimated at 80–90% after 2 days of incubation on potassium acetate medium. The other three lineages required 5–10 days on sporulation media. The West African strain gave usually 10–30% of sporulating cells. The Wine/European and Sake background strains yielded significantly fewer tetrads than the other strains, with 1–5% of sporulation. These results were consistent with the previous report (Cubillos et al., 2009).

Subsequently, a leu2∆0, lys2∆0, met15∆0 triple mutant in each of the five lineages was generated by successive crossings (Fig. 1B). First, two single mutants with distinct auxotrophies and opposite mating types were crossed to generate a diploid strain. From this diploid strain, spores were dissected to select colonies carrying double auxotrophies. A double-auxotroph haploid strain was then crossed with a haploid strain carrying the third single auxotrophy. Spores obtained from this diploid strain, carrying the leucine, lysine and methionine auxotrophies, were selected (leu2∆0, lys2∆0, met15∆0).

URA3 and KanMX cassette as additional selectable markers

The leu2∆0, lys2∆0, met15∆0 mutant haploid strains made in each of the five lineages carried a ura3∆::KanMX allele. In order to make complete markerless auxotroph strains which would also allow the recycling of the KanMX cassette for further genetic manipulations, we removed the KanMX resistance cassette that was used to inactivate the URA3 gene (Cubillos et al., 2009) and generated ura3∆0 mutant strains. In each of the five lineages, the MATa, ura3∆::KanMX, leu2∆0, lys2∆0, met15∆0 mutant strain was first transformed with a DNA fragment containing a wild-type copy of the URA3 gene to restore the uracil prototrophy phenotype (Fig. 1C). The transformation efficiencies varied from one strain to another and reached 38, 5, 48, 189 and 6 transformants/µg DNA amplicon for the Wine/European, West African, North American, Sake and Malaysian strains, respectively (Table 2).

The subsequent uracil prototroph strains were then transformed with a DNA carrying a deleted version of URA3 corresponding to the entire ORF from the start to the stop codon (Fig. 1C). The estimated transformation efficiency was 6 transformants/µg DNA for the Wine/European, West African and Sake strains, 2 transformants/µg DNA for the North American strain and 1 transformant/µg DNA for the Malaysian strain (Table 2). The quadruple auxotroph strains obtained here were leu2∆0, lys2∆0, met15∆0, ura3∆0.

A set of 50 auxotroph strains to explore S. cerevisiae natural variation

In an effort to eliminate putative secondary mutations that might have arisen during the genetic manipulations, the strain carrying all four deletions (MATa, leu2∆0, lys2∆0, met15∆0, ura3∆0) in each lineage was backcrossed twice with its respective parental strain (MATα, ura3∆::KanMX) (Fig. 1D). Spores resulting from these backcrosses and carrying eight different genotypes were collected for each lineage (Fig. 1E, Table 3). Auxotroph haploid strains for leucine and lysine or leucine and methionine were selected and then crossed to produce two distinct diploid strains with balanced gene markers, i.e. heterozygotes for the lys2∆0 and met15∆0 mutations (Fig. 1E, Table 3). A collection of 50 strains was assembled, including 40 haploid strains and 10 diploid strains (Table 3). Haploid strains that are mutated for the four biosynthesis pathways of lysine, leucine, methionine and uracil are available in each of the five lineages. These quadruple auxotroph strains will allow the use of most dominant genetic markers, including KanMX and NatMX (Goldstein and McCusker, 1999), with the exception of the HphMX cassette, which was used to delete the HO gene, which greatly enhances the possibilities for developing genetic systems. Furthermore, a version of each strain is available with the ura3∆0 mutation and the ura3∆::KanMX-barcode mutation labelled with a background-specific barcode, which is designed for identification and quantification purposes. In addition, both MATa and MATα strains were recovered, so that diploid derivatives could be made. The heterozygous diploid strains will also allow the generation of multiple derivatives by combining mutations as needed.

Table 3. Set of 50 auxotroph strains generated in this study
NameBackgroundMating typeura3Δ::KanMX-barcodeura3Δ0leu2Δ0lys2Δ0met15Δ0NCYC No.
  1. +, Deletion present; –, deletion absent.

  2. The HO gene was replaced by the HphMX cassette in all strains (hoΔ::HphMX in haploids and hoΔ::HphMX/hoΔ::HphMX in diploids).

YLF183West Africana++++3892
YLF175West Africana++++3893
YLF163West Africanα++++3894
YLF164West Africanα++++3895
YLF184West Africana+++3896
YLF176West Africana+++3897
YLF162West Africanα+++3898
YLF161West Africanα+++3899
YLF187West Africana/α−/−+/++/++/−+/−3900
YLF188West Africana/α+/+−/−+/++/−+/−3901
YLF130North Americana++++3902
YLF131North Americana++++3903
YLF148North Americanα++++3904
YLF149North Americanα++++3905
YLF133North Americana+++3906
YLF132North Americana+++3907
YLF147North Americanα+++3908
YLF146North Americanα+++3909
YLF190North Americana/α−/−+/++/++/−+/−3910
YLF189North Americana/α+/+−/−+/++/−+/−3911

Growth characteristics of the panel of auxotrophic strains

Besides differences in sporulation efficiency (see supporting information, Table S1), the five lineages displayed distinct levels of cell aggregation. For each lineage, we observed under the microscope that cells clump together to form aggregates that cannot be dissociated (either by sonication or by treatment with EDTA or mannose to inhibit lectin-mediated cell aggregation). This was particularly evident in the Malaysian lineage (UWOPS 03–461.4 strain). However, the other backgrounds also showed some levels of aggregation, which were significantly higher for the North American (YPS128) and Wine/European (DVBPG6765) lineages than for the West African (DVBPG6044) and Sake (Y12) strains. These characteristics should be taken into account in experimental designs where cell concentration must be precisely measured.

Further, growth of the 50 auxotroph strains generated in this work was measured in rich and complete synthetic liquid media. Growth curves allowed the calculation of generation times and growth rates for each strain and for their respective parents from the North American, Wine/European, West African and Sake groups (see also supporting information, Figure S1, Table S2). However, the high level of aggregation of the Malaysian strains prevented an accurate measurement of their growth (see below and supporting information, Figure S2).

In our hands, four of the five lineages and their mutant derivatives presented a typical-like growth, so that mutations did not appear to be deleterious to these four lineages. However, it should be noted that the four backgrounds behaved distinctively from each other and that, for instance, their growth curves had different features. In rich medium, the mutant derivatives behaved similarly to their respective parents for all four genetic backgrounds and for both haploid and diploid strains (see supporting information, Figure S1, Table S2). The mutant derivatives also behaved similarly to the S. cerevisiae BY4741 laboratory strain (see supporting information, Table S2). In complete synthetic medium, the growth of all West African mutant derivatives was similar to that of the parental strain (in both haploid and diploid contexts). However, the growth of the Wine/European, North American and Sake mutant haploid and diploid strains was slower than that of their respective parents, with a 1.5- and 1.8-fold increase of the generation time for the Wine/European haploid and diploid mutants, respectively, and a ~2- and 3-fold increase for the North American and Sake haploid and diploid mutants, respectively. In addition, the generation time of the diploid mutants in minimal medium is significantly higher than that of the corresponding haploid mutant in the Wine/European, North American and Sake groups; being most pronounced in the North American strain (see supporting information, Figure S1, Table S2).

The high level of aggregation of the Malaysian strains prevented the use of the TECAN plate reader to assess fitness. Therefore, dry weight experiments were performed, in both rich and minimal media, to compare growth between mutants and their respective parental strains. For both haploid and diploid mutant strains, no significant growth defect was observed in rich medium, while in minimal medium the mutant strains showed important growth defects relative to their parents (see supporting information, Figure S2).

Altogether, these data emphasize the fact that multiple auxotrophies generally alter cell growth rates in stringent culture conditions, such as when limited nutrients are present (Mulleder et al., 2012). A similar growth defect is also observed for the BY laboratory strains (both haploid and diploid) when cultivated in minimal medium (see supporting information, Table S2). It is nonetheless interesting to note that the West African mutant strains did not show any diminished growth in complete synthetic or rich media.

In conclusion, we envision that the use of this set of markerless auxotroph strains, which originate from highly diverged genetic backgrounds, will allow the development of new approaches to experimentally measure the role of natural variation and genetic interaction upon many biological functions and adaptation processes. Individual genome analysis alone (Jelier et al., 2011), or integrated with additional intermediate phenotypes datasets (Skelly et al., 2013), can be used to develop predictive genotype–phenotype models. It is also interesting to note that a recent multi-locus sequence analysis indicates the presence of eight highly diverged lineages from China (Wang et al., 2012), which vastly expand the known genetic variation of S. cerevisiae. An approach similar to the one described in this work could also be applied to these natural isolates (and also to additional lineages described in the future) to broaden our understanding of the relationship between genotypes and phenotypes.


We thank Jaime Hughes, Conrad Nieduszynski, Edward Louis, Héloïse Muller and Guy-Franck Richard for providing us with the plasmids and parental yeast strains used in this study. We are grateful to Natacha Sertour and Christophe d'Enfert for their help in performing the growth curve experiments. We thank Joseph Schacherer for providing us with the growth curve analysis tool. This work was supported by an ATIP grant from the Centre National de la Recherche Scientifique.