Synthetic two-species allodiploid and three-species allotetraploid Saccharomyces hybrids with euploid (complete) parental subgenomes

Combination of the genomes of Saccharomyces species has great potential for the construction of new industrial strains as well as for the study of the process of speciation. However, these species are reproductively isolated by a double sterility barrier. The first barrier is mainly due to the failure of the chromosomes to pair in allodiploid meiosis. The second barrier ensures that the hybrid remains sterile even after genome duplication, an event that can restore fertility in plant interspecies hybrids. The latter is attributable to the autodiploidisation of the allotetraploid meiosis that results in sterile allodiploid spores (return to the first barrier). Occasionally, mating-competent alloaneuploid spores arise by malsegregation of MAT-carrying chromosomes. These can mate with cells of a third species resulting in aneuploid zygotes having at least one incomplete subgenome. Here we report on the construction of euploid three-species hybrids by making use of “rare mating” between a sterile S. kudriavzevii x S. uvarum allodiploid hybrid and a diploid S. cerevisiae strain. The hybrids have allotetraploid 2nScnSk nSu genomes consisting of complete sets of parental chromosomes. This is the first report on the production of euploid three-species Saccharomyces hybrids by natural mating, without genetic manipulation. The hybrids provide possibilities for studying the interactions of three allospecific genomes and their orthologous genes present in the same cell.

The genus Saccharomyces comprises eight "natural species", namely S. arboricola, S. cerevisiae, S. eubayanus, S. jurei, S. kudriavzevii, S. mikatae, S. paradoxus, and S. uvarum (recently reviewed by Alsammar and Delneri 1 ) and many strains of chimeric (admixed) genomes that are, somewhat superficially, also called "interspecies hybrids". Two groups of the chimeric, mostly brewing strains of highly diverse genome structures are accommodated in the so-called "hybrid species" S. bayanus and S. pastorianus (S. carlsbergensis) (e.g. [1][2][3][4][5]. The chimeric strains identified in other environments (e.g. in wine-making processes) are not grouped in separate species and are assumed to have evolved from hybrids of natural species by loss and rearrangements of mosaics in the parental subgenomes (for a review, see e.g. 6 ).
The taxonomic division of the genus was mainly based on the biological species concept and later confirmed by the analysis of barcode and genome sequences. In the biological species concept introduced to Saccharomyces taxonomy by Naumov 7 , the species are populations of interbreeding strains isolated by sterility barriers. While conspecific strains form fertile hybrids (producing functional gametes), the strains that belong to different species either do not form hybrids (prezygotic sterility barrier) or their hybrids do not form functional gametes (ascospores) (postzygotic sterility barrier). The Saccharomyces species are isolated by postzygotic sterility barriers. All Saccharomyces species can form viable (allodiploid) hybrids with any other Saccharomyces species but the hybrids either do not sporulate or their spores are not viable.
Allodiploid sterility is mainly due to the failure of the chromosomes of the subgenomes to pair in meiosis I (e.g. [8][9][10][11] ) which results in the abruption of the meiotic process ("first sterility barrier"). In plants, allodiploid sterility can be circumvented by genome duplication, which "diploidises" the subgenomes (for a review, see 12

Methods
Strains and culture media. All strains used in this study are listed in Table 1. The medium used for the maintenance of the parental strains 10-170, 10-1651 and 10-1653 was YEA (yeast extract glucose agar). Hybrids were isolated and maintained on MMA (minimal medium agar) or on MMA supplemented with uracil. Mating tests were performed on YEA plates. Sporulation was tested on acetate SPA (sporulation agar) plates. Cultures for DNA isolation, karyotyping and flocculation tests were grown in YEL (YEA without agar). The composition of these media was described previously 24,25 . For FACS analysis, cells were propagated in YPD (YEL supplemented with 2% peptone).
Hybridisation. Hybridisation was based on complementation of auxotrophic markers. Two-species kudvarum hybrids were obtained by mating 10-1653 S. kudriavzevii with 10-1651 S. uvarum and selection of colonies growing on MMA supplemented with uracil. Since both parental strains were ura3 − , the two-species hybrids were auxotrophic for uracil. Their uracil auxotrophy was exploited at the construction of three-species cekudvarum hybrids. One of the kudvarum hybrids (II.6) was mated with 10-170 S. cerevisiae leu2 and prototrophic colonies were selected on MMA plates. Hybridisation was performed in two ways: by mass-mating of cells of exponential-phase cultures 19 and by a two-step replica-plating method 26 . Individual colonies (as products of individual zygotes) were isolated from the plates and stored at − 80 °C to prevent postzygotic changes in the genomes and segregation.  Table 3. For the amplification of the marker sequences, genomic DNA was isolated from 50-ml overnight cultures grown in YEL at 26 °C 24 . For the differentiation of the genes of the parental genomes, the amplified fragments were digested with restriction endonucleases that generated specific restriction patterns for each orthologue of each gene (Table 1S). The number and size of the subfragments generated by the digestion were determined by electrophoresis in 1.4% agarose gel, 0.5 × TBE. mtDNA extraction and RFLP. Mitochondrial DNA was prepared from exponential-phase YEL cultures with the method described by Nguyen et al. 28 and digested with MboI. The fragments were separated by gel electrophoresis in 0.7% agarose, 0.5 × TBE.
Mating and spore viability tests. Mating activity was tested in exponential-phase mixed cultures as described previously 19 . Briefly, equal volumes of overnight cultures of the strains grown in YEL were mixed, www.nature.com/scientificreports/ centrifuged and then 10 μl of the wet pellet was dropped on YEA. After incubation at room temperature for 4-6 h, samples were taken and examined microscopically. The testers of mating competences were the parental strains of the hybrids. Spore viability was examined by tetrad analysis. Samples of cultures grown on the sporulation medium SPA at room temperature for 5 days were suspended in Zymolyase-T20 (0.05 mg ml −1 ) solution. After incubation at 37 °C for 20 min, aliquots were streaked on YEA plates, and four-spored asci were pulled out from the streaks with a Carl Zeiss 2588 micromanipulator. The asci were dissected with the micromanipulator and the free spores were separated from each other on the plate to let the viable spores form individual colonies.
Physiological tests. Strains were tested for the utilisation of sugars as carbon sources in Durham tubes filled with YEL in which glucose was replaced with different sugars. The sensitivity of the strains to higher temperatures was compared by culturing them on YEA plates at 25 °C and 35 °C for 3 days. Their ability to flocculate was examined by culturing them in YEL on an orbital shaker at room temperature for 2 days. To visualise the aggregates, the cultures were poured into glass Petri dishes and photographed on dark background. Table 2. List of markers used for the identification of parental chromosomes in the hybrid genomes. 1 In reference 29 or in the Saccharomyces Genome Database (SGD): https:// www. yeast genome. org/. 2 Genes located on chromosomes that are not counterparts of the S. cerevisiae chromosomes are shown on bold font.

Results
Construction of sterile allodiploid and allotetraploid two-species and three-species hybrids. By mating double auxotrophic heterothallic S. kudriavzevii and S. uvarum strains, sterile urakudvarum hybrids were produced. Despite their sterility, the urakudvarum hybrids formed prototrophic threespecies cekudvarum hybrids at low frequency with the S. cerevisiae strain having complementary auxotrophy. The cekudvarum cells were also sterile. One kudvarum strain (II/6) and its 5 cekudvarum hybrids (II/6.1 to II/6.5) were chosen for further examination. II/6, II/6.1 and II/6.2 were used in a parallel project for the investigation of the role of the MAT locus in the yeast-specific second sterility barrier 19 but their genomes were not examined in detail. The genome size of the hybrids and their parental strains was compared by flow cytometry analysis (Fig. 1). The fluorescence peaks of the heterothallic parental strains 10-1651 S. uvarum and 10-1653 S. kudriavzevii had identical positions and could be attributed to cells being in G1 (1C amount of DNA) and G2 (2C amount of DNA) phases of the cell cycle. The positions of the S. cerevisiae culture indicated that its cells had 2C and 4C amount of DNA in the G1 and G2 phases. The increased genome size can be attributed to the instability of the heterothallism of this strain. It forms asci on the sporulation medium and rarely also on YEA. Sporulation indicates that it has become homothallic and its cells are diploid. The three-species hybrids which we produced in our previous study and analyse here might have arisen by rare mating between sterile allodiploid kudvarum cells and sterile diploid S. cerevisiae cells (allotetraploid three-species hybrid) or by "half-rare" mating between sterile kudvarum cells and fertile haploid S. cerevisiae ascospores (allotriploid three-species hybrid). However, the flow cytometry analysis measured tetraploid genomes (a peak located in a position corresponding to the 4C peak of the kudvarum parent and a peak behind it). Thus, the three-species cekudvarum hybrids had tetraploid amount of DNA. This result makes it unlikely that the hybrids were formed by "half-rare" mating.

The hybrids have alloeuploid karyotypes. Measuring of the DNA content of cells by flow cytometry
gives information about the ploidy, but provides no insight in the composition of the genome. It is not suitable for the investigation of the contribution of the parental genomes to the hybrid genome. The increased size of the latter can be due to the presences of complete parental subgenomes or to partially incomplete and partially duplicated subgenomes. Since the number of the chromosomes is identical in the three species used in this study but many of them differ in size 24,29 , the origin of most chromosomes of a hybrid can be inferred from their size. Therefore, we compared the karyotypes of the hybrids with those of the parental strains by pulsed-field gel electrophoresis. The karyotype of the two-species hybrid (II/6) shown in Fig. 2A contained all chromosomal bands of both parental strains. The number of bands further increased in the three-species karyotypes but certain chromosomes were not separated clearly. Neither the extension of the run-time of the electrophoresis nor the www.nature.com/scientificreports/ changes of the running parameters could separate them unambiguously. Although it could logically be supposed that the drastic increase in the number of bands was due to S. cerevisiae chromosomes, we wanted to prove this fact experimentally. Therefore we probed the gel with labelled S. cerevisiae-specific Y' telomeric sequences which only bind to the S. cerevisiae chromosomes 28 . As shown in Fig. 2B, all chromosomes of the S. cerevisiae parent and their size equivalents in the three-species genomes bound the probe. The single positive band in the other parental karyotypes can be attributed to non-specific binding. From the results of the flow cytometry and karyotype analyses it can be concluded that both types of hybrids had alloploid genomes.

PCR-RFLP analysis of marker genes verifies the euploidy of hybrids.
To confirm that the hybrids had euploid genomes, we tested them for the presence of parental orthologues of 34 "genetic markers" (genes and loci) that covered the entire chromosomal sets of the parental strains. The orthologues of the markers could be differentiated by PCR-RFLP due to their different restriction patterns (Supplementary Table S1 and examples in Fig. 3 and Supplementary Fig. 2S). Since the genomes of the three species are not entirely syntenic, 11 markers were located on different (non-homeologous) chromosomes (marked with grey in Table 2) in their genomes. Therefore at least two markers were chosen for each of the 16 chromosomes, in most cases from different arms.
In the case of three markers GND1, CYR1 and MET2 (located on Chr VII, X and XIV of S. cerevisiae, respectively) the restriction patterns did not differ sufficiently for distinguishing all three orthologous (Table S1). The bands of the S. kudriavzevii pattern of OPY1 (Chr II in S. cerevisiae) were not visible in the hybrids. Since other markers of these chromosomes showed different parental patterns, these chromosomes could also be detected. The PCR-RFLP analysis identified all S. kudriavzevii and S. uvarum chromosomes in the two-species kudvarum hybrid II/6. The three-species cekudvarum hybrids also had all S. cerevisiae chromosomes. Taking all PCR-RFLP results together, it can be concluded that both the two-species and the three-species hybrids had complete sets of parental chromosomes.

The mitochondrial genome is inherited uniparentally. Digestion of the isolated mitochondrial DNA
with MboI generated different band patterns for the parental strains (Fig. 4). The pattern of the two-species hybrid II/6 was identical with that of the S. kudriavzevii parent. The 5 three-species hybrids had identical mito- www.nature.com/scientificreports/ chondrial genomes whose MboI patterns were indistiguishable from that of the S. cerevisiae parent. Thus, the hybrids were homoplasmic and the mitochondrial genomes were inherited uniparentally.

Dominant/recessive relationships in the determination of phenotypic traits.
Since the species used for hybridisation differ in certain taxonomically relevant phenotypic traits, we tested the hybrids for these properties. The growth of S. uvarum is inhibited by temperatures above 35 °C, whereas the other species can grow at these temperatures. Neither the kudvarum nor the cekudvarum hybrids were sensitive to 35 °C, so the temperature sensitivity of S. uvarum is recessive (Supplementary Fig. 3S). S. uvarum can utilise melibiose as a carbon source whereas the other species are mel − . Both types of hybrids grew in the medium in which glucose was replaced with melibiose and also could ferment it. Thus, this trait of S. uvarum was dominant. S. uvarum and S. kudriavzevii also differ in maltose and galactose utilisation (S. uvarum is mal + and gal + ) and flocculation of cells (S. kudriavzevii is highly flocculant). All hybrids were able to utilise both carbon sources, indicating that these traits are also determined by dominant alleles. The genetic determination of flocculation appears to be more complex. As shown in Fig. 5, II/6 flocculated like the S. kudriavzevii parental strain but the cekuvarum hybrids did not flocculate.

Discussion
In a previous study we created two-species kudvarum and three-species cekudvarum hybrids to investigate the role of the MAT locus in the postzygotic sterility barriers that biologically isolate the Saccharomyces species from each other 19 . Here additional hybrids were produced and the genome structures of selected representatives were investigated. As expected on the basis of numerous previous observations (reviewed e.g. in Reference 6 ), the twospecies hybrids were sterile. In many plants, the sterility of interspecies hybrids can be overcome by genome duplication (e.g. 12 ). Saccharomyces allodiploid hybrids can also duplicate their genomes but the duplication does not restore fertility because the yeast allotetraploids do not form functional (mating-competent) gametes. However, occasional imprecise partitioning of chromosomes during allotetraploid meiosis can result in matingcompetent spores. The spore receiving only one MAT-carrying chromosome (loss of MAT heterozygosity) can conjugate with other spores or cells 14,19 . The regained fertility allows hybridisation with a third species but the hybrids will not have complete parental genomes because the lost chromosome(s) will be missing 21,22 . Since we wanted to create three-species hybrids possessing euploid genomes (complete parental subgenomes), we opted for a different hybridisation strategy. We made use of the rarely occurring "escape" from the repression of the mating programme by the MAT heterozygosity "are mating" 23 ). Although rare mating was originally observed in S. cerevisiae autodiploids, we found in this study that mating-competent cells also occur in kudvarum allodiploid cultures that can mate with S. cerevisiae cells to form three-species cekudvarum hybrids. www.nature.com/scientificreports/ The flow cytometry analysis determined 2C and 4C amounts of DNA in the kudvarum and cekudvarum hybrids, respectively. Since the S. kudriavzevii and S. uvarum strains were stable heterothallic haploids and the S. cerevisiae was diploid, we inferred from the flow cytometry results that the kudvarum hybrids had allodiploid n Sk n Su genomes and the cekudvarum hybrids had allotetraploid 2n Sc n Sk n Su genomes. In the electrophoretic karyotypes, the hybrids had equivalents of all chromosomal bands of the parents.
However, neither flow cytometry analysis nor karyotyping can unambiguously prove that the hybrids have complete (euploid) subgenomes. The FACS analysis is not sufficiently sensitive to detect differences in DNA content arising from loss or duplication of single chromosomes, and karyotyping cannot separate chromosomes similar in size. To identify each chromosome individually, we tested the hybrids for the presence of orthologues of a group of selected genes as chromosome-specific molecular markers that covered all chromosomes of all parental strains. The RFLP analysis of these markers identified complete sets of S. kudriavzevii and S. uvarum chromosomes in the two-species kudvarum hybrids, and the three-species cekudvarum hybrids also had all S. cerevisiae chromosomes. Therefore, both types of hybrids had euploid genomes.
Neither the kudvarum nor the cekudvarum hybrids formed viable spores. The failure of interspecies allodiploid hybrids to produce viable gametes can be attributed to the failure of the allosyndetic (homeologous) chromosomes of their subgenomes to pair during prophase I of meiosis (e.g. [8][9][10][11] ). Even if the homeologous chromosomes are syntenic enough for aligning with each other, their sequence differences prevent them from efficient DNA strand exchange necessary for pairing up in Prophase I 11 . Since both types of hybrids had single copies of S. kudriavzevii and S. uvarum chromosomes, normal meiosis could not take place and viable gametes could not be produced. The presence of two sets of S. cerevisiae chromosomes did not improve the situation despite the possibility of normal pairing within the S. cerevisiae subgenome. Previous studies have shown that when the alloploid (e.g. allotetraploid) hybrid had eudiploid subgenomes, the chromosomes paired preferentially with their homologues within the autodiploid subgenomes. This mode of meiosis, referred to as autodiploidised allopolyploid meiosis, produces viable spores 15 . The cekudvarum hybrids could not form viable spores because only one of the subgenomes was autodiploid.
Since no mitochondrial markers were used in the construction of hybrids, the transfer of the mitochondria from the parental cells to the hybrids did not take place under selection pressure. In such circumstances, the mitochondria of both mating partners can be transmitted into the zygote. However, heteroplasmic interspecies hybrids were rarely observed when different Saccharomyces species were hybridised in previous studies. The hybrids usually had parental mitotypes or, less frequently, recombinant mitotypes (e.g. 16,[30][31][32][33][34]. In this study both types of hybrids were homoplasmic. The two-species kudvarum hybrids received their mtDNA from S. kudriavzevii. This was then replaced with the mtDNA of S. cerevisiae in the three-species cekudvarum hybrids. In both cases the mitochondrial genome was inherited uniparentally. In previous studies, we also observed uniparental inheritance of S. cerevisiae mitochondrial genome in cevarum (S. cerevisiae x S. uvarum) hybrids 16,24 . www.nature.com/scientificreports/ Three-species hybrids provide possibilities to study the interactions of three orthologues (alleles) of genes within one strain. In a previous paper we found that the genes of the MAT loci and the HO genes of three subgenomes cooperated in the hybrids as efficiently as their counterparts in the parental strains 19 . Here we show that the temperature sensitivity of S. uvarum is recessive both in the two-species and in the three-species hybrids, whereas the ability of this species to utilise galactose, maltose and mellibiose as carbon sources is dominant. The relationships of the determinants of flocculation appear to be more complex: this trait characteristic of the S. kudvarum cells was dominant in kudvarum but recessive in the cekudvarum hybrids.
The results presented in this study demonstrate that three-species euploid hybrids can be constructed by making use of natural mating processes and complementation of auxotrophic phenotypes without the application of genetic engineering. These hybrids allow the investigation of interactions of complete gene pools of three species, subsets of genes involved in complex physiological properties and individual groups of orthologues. Being non-GMOs, these hybrids and their segregants formed by postzygotic evolution of their genomes (e.g. by GARMi and GARMe) can be exploited in biotechnological processes even in countries whose legislations restrict or prohibit the use of genetically modified organisms.

Data availability
All data generated or analysed during this study are included in this published article and its supplementary information files.