Strains of Saccharomyces cerevisiae, the major yeast used in fermentation technologies (Tuite and Oliver, 1991) show high genetic and phenotypic diversity. The favorable properties of the different strains can be brought together by hybridisation. The hybrids not only show transgressive phenotypes but also produce mitotic and meiotic segregants of diverse combinations of the traits of the hybridizing partners. Strains of the other species of the genus can have additional favorable properties but their combination with those of the S. cerevisiae strains is hampered by the postzygotic reproductive isolation of the species manifested as hybrid sterility. The interspecies hybrids of the Saccharomyces species are viable but either do not produce gametes (ascospores) or if they do so, the viability of the gametes is extremely low. This sterility barrier keeps the species biologically isolated but the isolation is not absolute. The hybrid genomes can change and certain types of changes make the barrier permeable. In a previous review a model was proposed to integrate these postzygotic events into a coherent system based on what was then known (Sipiczki, 2008). According to the model, the hybrid genome undergoes a gradual size reduction by losing chromosomes, either in the course of vegetative propagation of the allodiploid hybrid cells or during allotetraploid meiosis which takes place upon spontaneous genome duplication. Concomitantly with size reduction the subgenomes can interact and recombine. The “stabilized” outcomes of these processes are recombinant haploids and aneuploids, in fact strains with chimeric (mosaic) genomes. Hybridisation and postzygotic genome chimerisation can be observed in the laboratory but can take place also in natural habitats as demonstrated by the occurrence of chimerised genomes in strains isolated from yeast communities fermenting beverages. Over the past 10 years, considerable progress has been made in the investigation of hybrid sterility, the breakdown of the sterility barrier, and the mechanisms underlying the postzygotic reduction and chimerisation of the hybrid genome. These processes and their exploitation in the improvement of industrial Saccharomyces strains, as a non-GMO alternative of targeted genetic manipulation, are the subjects of this review. A review of this length cannot be comprehensive and thus it will not cover the hybrid species, the natural “hybrid strains” and the evolutionary aspects of hybridisation. The reader interested in the developments in these fields can consult review papers (e.g., Sipiczki, 2008; Louis, 2011; Albertin and Marullo, 2012; Morales and Dujon, 2012; Dujon and Louis, 2017; Krogerus et al., 2017a; Bisson, 2017; Gibson et al., 2017; Guillamón and Barrio, 2017; Lopandic, 2018) published elsewhere. Given that certain genetic terms are often inconsistently used in the literature, a section will address terminological issues.

The taxonomy of Saccharomyces changed many times in the history of the genus. van der Walt (1970) separated the highly fermenting species from the rest of the genus and proposed the name Saccharomyces sensu stricto complex for them. Since then the species not included in this group (sensu lato) were transferred to other genera, so the name “sensu stricto complex” has become obsolete. Yeast taxonomy currently accepts 7 “natural,” “clean” or “single-genome-based” [S. arboricolus (S. arboricola), S. cariocanus, S. cerevisiae, S. kudriavzevii, S. mikatae, S. paradoxus, and S. uvarum (S. bayanus var. uvarum)] and 2 “hybrid” [S. pastorianus/carlsbergensis and S. bayanus (S. bayanus var. bayanus)] Saccharomyces species (Vaughan-Martini and Martini, 2011). Since 2011 two new Saccharomyces species S. eubayanus and S. jurei were described. However, this classification is in contradiction with the results of whole-genome sequencing. Whole-genome analysis has shown that Saccharomyces cariocanus should be considered to be more a S. paradoxus variant as this species is reproductively isolated from Saccharomyces paradoxus by four translocations but not by sequence (reviewed in Borneman and Pretorius, 2015; Dujon and Louis, 2017; Nguyen and Boekhout, 2017). But it does not fit with mtDNA gene order, which is considered as a species-specific feature. The S. cariocanus mtDNA is not syntenic to that of S. paradoxus and therefore S. cariocanus should be designated as a separate species (Sulo et al., 2017).

The hybrid species (S. bayanus, and S. pastorianus/ carlsbergensis) occurring almost exclusively in beer fermentation are highly heterogeneous in genome structure and the proportion of the genomic mosaics originating from the assumed parental species (e.g., Nguyen and Gaillardin, 2005; Rainieri et al., 2006; Dunn and Sherlock, 2008; Nakao et al., 2009; Libkind et al., 2011; Nguyen et al., 2011; Walther et al., 2014; Pérez-Través et al., 2014a; Van den Broek et al., 2015). In addition to these hybrid species, many other Saccharomyces strains have been identified mainly among wine yeasts that contained mosaics of foreign origin in their genomes. These strains usually referred to as “hybrids” are not considered distinct taxonomic entities (for reviews, see Sipiczki, 2008; Albertin and Marullo, 2012; Morales and Dujon, 2012; Dujon and Louis, 2017; Lopandic, 2018). There is no consensus about where to draw the line between the hybrid species and the interspecies hybrids. Both groups have diverse chimeric genomes.

There is considerable confusion also around the taxonomic identity of S. bayanus and S. uvarum. Both species have been described in the 19th century but because of their proposed merger under the name S. bayanus some three decades ago (Vaughan-Martini and Kurtzman, 1985), the name S. uvarum has become sparsely used in the literature until recently. However, the molecular genetic and genomic analyses clearly separated two groups, a clean and a hybrid group, among the strains that had been assigned to S. bayanus over the last 30 years and led to the reinstatement of the species S. uvarum for the clean group (Nguyen and Gaillardin, 2005; Nguyen et al., 2011). As most S. bayanus strains isolated from substrates not related to beer production have turned out in molecular tests to be conspecific with the type strain of S. uvarum, the name S. bayanus will be restricted only to brewery yeasts in this review.

All discrepancies in fungal taxonomy come from the three species concepts that have been applied to species delimitation: the classical morphology/physiology-based, the biological and the evolutionary/phylogenetic species concepts. The classical concept is difficult to apply generally in Saccharomyces taxonomy because of the strain variation in the utilization of different substrates and the absence of distinctive morphological properties. The biological species concept, which defines a species as a group of actually or potentially inbreeding natural population reproductively isolated from other such groups (Mayr, 1940), was introduced to Saccharomyces taxonomy by Naumov (1987) and successfully applied to the description of novel species and to the determination of the taxonomic affiliation of novel isolates. This approach is based on the observation that the hybrids of the Saccharomyces strains are viable, but only hybrids of conspecific strains can produce viable spores at frequencies higher than 1%. However, this method is impractical in large-scale screening and routine testing of taxonomic affiliation of larger numbers of isolates. The phylogenetic concept defines a species as a monophyletic group of organisms sharing molecular characters that derive from a common ancestor (Moore et al., 2011). The phylogenetic relationships of yeast strains can be examined by comparing the sequences of their evolutionary conserved genes or non-coding chromosomal segments such as the D1/D2 domains of the genes encoding the large-subunit (LSU) rRNA molecules and the internal transcribed segments (ITS), two components of the repeats of the rDNA arrays. A recent study of 9,000 yeast isolates of the CBS collection (Vu et al., 2016) found that strains differing by more than 0.49% in their D1/D2 domains and by more than 1.59% in their ITS segments are usually not conspecific. The “natural” Saccharomyces species can be separated with these criteria. The “hybrid species” S. pastorianus/carlsbergensis and S. bayanus are difficult to distinguish from the “natural species” because certain their strains show no sexual activity (untestable for biological isolation) and/or have rDNA sequences of different “natural” species in heterozygous constitution (e.g., Nguyen and Gaillardin, 2005; Rainieri et al., 2006; Nguyen and Boekhout, 2017). Nevertheless their type strains sporulate and it is possible to mate them e.g., to S. cerevisiae (Špírek et al., 2014). The pan-genome analysis (e.g., Tettelin et al., 2005; Dunn et al., 2012) of large numbers of strains of all species of the genus may shed new light on the problem of species boundaries.

Strains of the natural (“clean” or “single-genome-based”) Saccharomyces species are usually homothallic (for a review, see Rainieri et al., 2003). Their genomes are diploid or aneuploid and heterozygous for the mating-type alleles. MATa/MATalpha heterozygosity makes the cells unable to conjugate by repressing the mating program. However, it allows meiosis and sporulation in response to proper external signals (e.g., starvation). Meiosis produces haploid meiospores that have only one MAT locus, either MATa or MATalpha. The loss of MAT heterozygosity blocks the meiotic program and concomitantly abolishes the block of the mating program. So the spores can mate with other mating-competent conspecific spores and thus can act as gametes to form autodiploid zygotes. The zygotes then produce mitotically propagating autodiploid cells, equivalents of the somatic cells of higher eukaryotes. Having a MATa/MATalpha genotype, these cells are also sporogenic and can produce viable gametes (ascospores). For a review of the genetic determination of sexual processes of Saccharomyces strains, see Herskowitz (1988).

Mating can take place also between non-conspecific Saccharomyces spores. The resulting allodiploid zygotes are viable and produce vegetative cells much like the autodiploid zygotes. But in contrast to those, the allodiploid cells cannot sporulate or if they do so, their spores are very rarely viable (e.g., Gjermansen and Sigsgaard, 1981; Zambonelli et al., 1993; Hawthorne and Philippsen, 1994; Kishimoto, 1994; Hunter et al., 1996; Marinoni et al., 1999; Liti et al., 2006; Kunicka-Styczyńska and Rajkowska, 2011; Pfliegler et al., 2012). This phenomenon called hybrid sterility or hybrid incompatibility plays an essential role in the biological isolation of the Saccharomyces species because it impairs the competitiveness of the hybrids under unfavorable natural conditions (for explanation, see section the bad and good sides of sterility) and prevents the recombination of their gene pools.

What is the genetic basis of hybrid sterility in Saccharomyces?

Four major hybrid sterility mechanisms have been proposed:

In plants, the sterile allodiploid hybrids become fertile when they duplicate their genomes because in the allotetraploid genomes each chromosome has a homologous partner to pair with, and thus a normal meiosis can take place. Upon genome duplication the hybrids produce viable and functional allodiploid gametes (for a review, see Ramsey and Schemske, 2002). A seemingly similar process takes place also in Saccharomyces interspecies hybrids but with a different outcome. If the hybridisation of two Saccharomyces species is followed by genome doubling or the mating cells were diploid (rare event), the hybrid becomes able to form viable spores (Cummings and Fogel, 1978; Banno and Kaneko, 1989; Marinoni et al., 1999; Johnston et al., 2000; Naumov et al., 2000; Greig et al., 2002a; Sebastiani et al., 2002; Antunovics et al., 2005; Pfliegler et al., 2012; Karanyicz et al., 2017). This observation led to the widespread view that genome duplication restores fertility in Saccharomyces interspecies hybrids as well. However, where investigated, these spores turned out to possess allodiploid chromosomal sets and thus represented a return to the ancestral sterile allodiploidy (Sebastiani et al., 2002; Pfliegler et al., 2012; Karanyicz et al., 2017) (Figure 1). Their sterility makes them different from the gametes of the allotetraploid plants which are fertile. Another difference from the plant allodiploid gametes is that the yeast allodiploid gametes (F1 spores) can propagate vegetatively and produce allodiploid F1 spore clones. The third difference is that the cells of these clones can sporulate. But like the spores of the allodiploid hybrids, the spores of the allodiploid F1 spore clones are dead. The inability of the F1 spore clones of the allotetraploid cells to mate and produce viable spores is the second sterility barrier in the biological isolation of the Saccharomyces species (Pfliegler et al., 2012). Thus, in contrast to the plant interspecies hybrids, genome duplication does not restore fertility in the Saccharomyces hybrids, it only produces non-functional gametes that can propagate vegetatively but cannot fertilize (conjugate).

The pairing of a chromosome with its homologous partner (autosyndesis) in the allotetraploid meiosis excludes its pairing with a chromosome from the other subgenome (allosyndesis) (Ramsey and Schemske, 2002; Moore, 2002). Thus, each diploid subgenome performs meiosis without interaction with the companion subgenome and then transmits its halved (haploid) sets of chromosomes into each of the four gametes. This mode of meiotic division is referred to as autodiploidisation of the allotetraploid genome in plant genetics (Hutchinson et al., 1983). The investigation of the viable spores produced by allotetraploid Saccharomyces hybrids revealed that the meiosis of the allotetraploid yeasts is also autodiploidised (Pfliegler et al., 2012; Karanyicz et al., 2017). Like the gametes of the allotetraploid plants, the spores of the allotetraploid yeasts are allodiploid with a single set of chromosomes in both subgenomes. Each subgenome contains a MAT-carrying chromosome but their MAT alleles are different because the diploid subgenomes of the allotetraploid were homozygous for different alleles: one was MATa/MATa the other was MATalpha/MATalpha (because hybridisation can take place between cells of opposite mating types and mating-type switching is then repressed by the MAT heterozygosity; Herskowitz, 1988). The simultaneous presence of both MAT alleles makes the spores (and their vegetative progeny) of the allotetraploid hybrid unable to mate but able to launch the meiotic program, although the spores produced are not viable (Pfliegler et al., 2012). This is exactly what characterizes the allodiploid hybrids. Thus, after genome duplication and a round of successful meiosis the hybrid returns to the sterile allodiploid state (Figure 1). This mode of meiosis and chromosomes segregation is very different from those characteristic of autotetraploids. In an autotetraploid S. cerevisiae cell any chromosome can freely pair with any of the other three homologous copies regardless of their origin. As a result, many allodiploid spores will be homozygous for the mating type (MATa/MATa or MATalpha/MATalpha) and capable of mating. Neither the heterozygous spores are sterile because their vegetative descendants produce viable haploid spores (e.g., Hilger, 1973; Al Safadi et al., 2010).

The allotetraploid meiosis turned out to be prone to errors at the distribution of the chromosomes to the spores. It was noticed that certain asci of certain allotetraploids contained two types of spores, usually in 2:2 proportion. Two spores were sterile, and two spores formed clones of cells producing viable spores (Antunovics et al., 2005; Pfliegler et al., 2012; Karanyicz et al., 2017). The latter type lacked the MAT-carrying chromosome in one of the subgenomes. This chromosome (designated III in S. cerevisiae and III, 3 or 2 in other species, depending on the numbering system used) is known to be the least stable chromosome even in autoploid S. cerevisiae strains (Kumaran et al., 2013). Having only one MAT allele (being hemizygous at the MAT locus), the cells of these clones can (in contrast to the clones of the MATa/MATalpha allodiploid spores) switch their mating types and then conjugate with each other to form uniparentally disomic allotetraploid zygotes. This can take place because the loss of mating-type heterozygosity reactivates both the mating-type switching machinery and the mating program (Herskowitz, 1988). The intraclonal conjugation (selfing, autofertilisation) converts the F1 spore clone to an F2 population of cells having allotetraploid genomes nullisomic for the lost MAT-carrying chromosome of one subgenome and disomic for the MAT-carrying chromosome of the other subgenome. In these cells autosyndetic (homologous) chromosome pairing can take place, and thus the meiosis produces fertile alloaneuploid F2 spores. The cells of the F2 spore clones then form a fertile F3 generation and so on. These observations demonstrate that the loss of MAT heterozygosity by the loss of the MAT-carrying chromosomes in one of the subgenomes abolishes the sterility of the alloploid hybrid (Figure 2). The conjugation-proficient spore clones of the tetraploid S. cerevisiae × S eubayanus hybrids used by Krogerus et al. (2017b) for hybridisation with a S. cerevisiae strain might have been this sort of fertile alloaneuploids. The cost of the breakdown of the sterility barrier is the destabilization of the hybrid genome. The loss of the MAT-carrying chromosome converts the euploid genome into aneuploid after which additional chromosomes can be lost in rapid succession (see next section).

In principle, conjugation-proficient spores can occur also among the few viable aneuploid spores produced by allodiploid hybrids provided they have only one MAT-carrying chromosome. The fertile “haploid” spores described in Greig et al. (2002b) and in Lee et al. (2008) could have been such segregants but because their ploidy was not examined, the possibility cannot be excluded that they were allodiploid spores nullisomic for one parental MAT-carrying chromosome produced by cells of duplicated (allotetraploid) genomes.

The fertile alloaneuploid spores produced after the loss of MAT heterozygosity can conjugate with any fertile gamete of any other strain, also with the spores of one or the other parental species which opens the way for introgression. The caveat is that the products of these backcrosses will have low fertility due to their unbalanced sets of chromosomes and MAT heterozygosity. In an early work Sebastiani et al. (2002) observed conjugation between a few F1 spore clones of allotetraploids and the parental strains, but the hybrids obtained from the backcross produced mainly dead spores. de Barros Lopes et al. (2002) raised the possibility that introgression might take place by serial matings of the vegetative cells of the sterile hybrid with cells of one of the parental species [by the process called rare mating (Gunge and Nakatomi, 1972)]. Lee et al. (2008) managed to backcross the vegetative and autodiploidised (via self-fertilization) descendants of the spores of a S. cerevisiae × S bayanus (uvarum?) hybrid to the S. cerevisiae parent several times successively. Because of the very concise description of certain parts of the experimental procedure, it is impossible to resolve the inconsistency between the assumable MATa/MATalpha genotype (mating program repressed) of the cells and their ability to mate. It is conceivable that the spore clones contained conjugation-proficient segregants that had lost one of the parental MAT-carrying chromosomes. Nevertheless, the transfer of genetic information between Saccharomyces species via introgressive backcrosses has not been verified yet experimentally.

A recently described alternative process leading to chimerised genomes is GARMe (Genome Autoreduction in Meiosis) (Karanyicz et al., 2017) (Figure 3). It is a progressive process starting with the breakdown of the sterility barrier by the malsegregation of the MAT-carrying chromosomes during the allotetraploid meiosis. The loss of a pair of MAT-carrying chromosomes destabilizes the genome and additional pairs of autosyndetically paired chromosomes can malsegregate in the meiotic divisions of the consecutive filial generations of autofertilised spore clones. The genome gradually becomes smaller at each meiotic division, with one partner’s chromosomes being lost preferentially. Occasionally, recombination between chromosomes of the subgenomes can also take place (Antunovics et al., 2005; Karanyicz et al., 2017; Krogerus et al., 2017b), most probably due to allosyndetic interactions between similar segments of homeologous chromosomes (Lorenz et al., 2002). During GARMe, the hybrid genome is gradually transformed into various genomic chimeras comprising predominantly chromosomes of one species but containing also mosaics from the genome of the other species (Antunovics et al., 2005; Pfliegler et al., 2012, 2014; Lopandic et al., 2016; Karanyicz et al., 2017). The process gradually leads to chimeric genomes similar to those found in “natural hybrids” (in fact chimeras) that have only a few genes from the donor genome (see above). The differences between the hybrids created by hybridizing conjugation-proficient spore clones of an allotetraploid S. cerevisiae × S. eubayanus hybrid and a S. cerevisiae strain (Krogerus et al., 2017b) might be attributed to different chromosome losses in the spores.

Vegetatively propagating diploid S. cerevisiae cells can lose chromosomes by rare spontaneous chromosome non-disjunction during mitosis. The event generates two aneuploid daughter cells one of which is monosomic, the other being trisomic (Parry and Zimmerman, 1976). Aneuploidy destabilizes the genome and thus initiates further karyotype changes (e.g., Parry and Cox, 1970; Zhu et al., 2012). As predicted in the model proposed a decade ago (Sipiczki, 2008), malsegregation leading to aneuploidy can occur also in vegetatively propagating populations of allodiploid hybrid cells. Recent studies confirmed the prediction. It was found that alloploid Saccharomyces hybrids can lose chromosomes during vegetative propagation, which can also be accompanied by other forms of structural rearrangement in the genome (e.g., Kunicka-Styczyńska and Rajkowska, 2011; Piotrowski et al., 2012; Dunn et al., 2013; Bellon et al., 2013, 2015, 2018; Pérez-Través et al., 2014b; Mertens et al., 2015; Lopandic et al., 2016; Krogerus et al., 2017b; Peris et al., 2017). Such derivatives (frequently called “evolved hybrids”) constitute minor components of the population of the hybrid cells but can become relatively more abundant if the hybrid culture is cultivated under enriching conditions favoring their propagation for longer periods of time (for a recent review, see Lopandic, 2018). Various “enriching” or selective conditions have been applied to obtain strains with specific properties such as anaerobiosis (Kunicka-Styczyńska and Rajkowska, 2011), high temperature and alcohol content (Piotrowski et al., 2012), ammonium limitation (Dunn et al., 2013), glucose limitation, phosphate limitation and sulfate limitation (Smukowski Heil et al., 2017) sulfate limitation (Sanchez et al., 2017), grape must fermentation (Pérez-Través et al., 2012, 2014b; Bellon et al., 2013, 2018; Origone et al., 2018), fermentation at high sugar concentrations (Lopandic et al., 2016), cultivation in lager beer medium (Mertens et al., 2015), in xylose fermentation medium (Peris et al., 2017), etc.

The most frequently detected genome modifications were losses of chromosomes and changes of the genome size. An extreme example is the loss of the almost entire S. uvarum subgenome during culturing of the S. cerevisiae x S. uvarum hybrid at increased temperature (31–46.5°C) (Piotrowski et al., 2012). Preferential alterations and reduction of one subgenome was reported several times. For example Bellon et al. (2018) also detected chromosomal loss and much more rearrangements in the S. uvarum subgenome than in the S. cerevisiae subgenome during must fermentation. Lopandic et al. (2016) observed chromosome loss in the S. uvarum subgenome of two F1 spore clones of a S. cerevisiae x S. uvarum allotetraploid and in the S. kudriavzevii subgenome of F1 spore clones of a S. cerevisiae x S. kudriavzevii allotetraploid hybrid during experimental wine fermentation.

In addition to aneuploidisation of the genome, also size changes of certain chromosomes were observed (e.g., Kunicka-Styczyńska and Rajkowska, 2011; Lopandic et al., 2016; Bellon et al., 2018) which indicates that various forms of recombination (e.g., exchange of chromosomal arms, translocation) can occur too. Other experiments demonstrated that recombination can take place also at the level of individual genes. Dunn et al. (2013) detected frequent recombination events between the MEP2 genes of the partner genomes in vegetatively propagating S. cerevisiae × S. uvarum hybrids. The observed genomic changes and interactions between the subgenomes in vegetatively propagating hybrids strongly suggest that the hybrid genome can be reduced and chimerised not only by GARMe but also during vegetative propagation without introgressive backcrosses. To the analogy of GARMe, this process can be called GARMi (Genome Autoreduction in Mitosis).

Different selective conditions can enrich derivatives of different phenotypes from the same hybrid (e.g., Piotrowski et al., 2012). When there is no selective pressure, the culture can segregate into subpopulations differing in genome size and chromosomal constitution even in the same culturing conditions (e.g., Mertens et al., 2015). The phenotypic diversity found in a recent study among hybrids obtained by hybridisation of the same pair of isogenic strains (Verspohl et al., 2018) might have been the consequence of different spontaneous genome rearrangements in the individual hybrid lines during their maintenance in non-selective culturing media.

The overgrowth of the unchanged hybrid cells by a segregant is a sort of adaptive evolution at population level because the cells of the segregant (evolved hybrid) are more fit than the original hybrid under the applied culturing conditions. The winner then remains stable, thus the process can also be regarded genome stabilization (e.g., Piotrowski et al., 2012; Pérez-Través et al., 2014b; Krogerus et al., 2017b; Bellon et al., 2018) but cultivation of the “stabilized” strain under different conditions would show whether its genome is stable indeed.

The aneuploid spore clones produced after the breakdown of the sterility barrier also change during vegetative propagation. Lopandic et al. (2016) found extensive segregation in F1 spore clones of S. cerevisiae × S. uvarum and S. cerevisiae × S. kudriavzevii hybrids, whereas Krogerus et al. (2017b) observed less drastic changes in S. cerevisiae × S. eubayanus hybrids. In both cases the non-cerevisiae subgenome was less stable.

When interpreting the changes of the hybrid genome during vegetative propagation, it has to be borne in mind that occasional meiotic divisions may also contribute to the process despite the adverse cultivation conditions for sporulation. Mortimer et al. (1994) observed sporulation in certain wine yeast strains even in sugar-rich media and Sebastiani et al. (2002) found that S. cerevisiae × S. uvarum hybrids could form asci on YPD (yeast-peptone-glucose medium).

In principle, GARMi can also break down the sterility barrier, if the hybrid cell loses the MAT-carrying chromosome in one of the subgenomes (loss of MAT heterozygosity). Kunicka-Styczyńska and Rajkowska (2011) found fertile segregants in cultures of sterile S. cerevisiae × S. uvarum hybrids after 50–80 generations of mitotically dividing cells.

As shown above, one of the subgenomes usually undergoes a faster and more extensive reduction both in the meiotic (GARMe) and in the mitotic (GARMi) divisions. The less stable subgenome in meiosis was that of S. uvarum in the S. cerevisiae × S. uvarum hybrids (Antunovics et al., 2005; Pfliegler et al., 2012; Piotrowski et al., 2012), S. eubayanus in the S. cerevisiae × S. eubayanus hybrids (Krogerus et al., 2017b), S. kudriavzevii in the S. cerevisiae × S. kudriavzevii (Pfliegler et al., 2014; Lopandic et al., 2016) and S. kudriavzevii × S. uvarum hybrids (Karanyicz et al., 2017). The mechanisms underlying the unequal reduction of the subgenomes (concerted loss of chromosomes in one of the subgenomes) are unknown but a hypothetical model was recently proposed to explain the asymmetry (Karanyicz et al., 2017). The model is based on the assumption that the loss of a chromosome is unlikely to affect physically the segregation of other chromosomes. More plausibly, it creates a situation in which the subsequent loss of certain chromosomes of the other subgenome would have deleterious effects. For example, when the fertility is restored by the loss of the MAT-carrying chromosome of one of the subgenomes, all genes located on that chromosome are lost. The absence of these genes has no discernible effect on viability because their orthologs in the companion subgenome can substitute them. Most of the orthologs are in the MAT-carrying chromosome of the partner subgenome but some of them are on different, non-homeologous chromosomes [e.g., as shown in S. kudriavzevii × S. uvarum hybrids (Karanyicz et al., 2017)]. If one of these non-homeologous chromosomes is then lost during the next meiosis, the orthologues located on it will be eliminated from the genome. If any of them performs an essential function in the life of the cell, the spore lacking this chromosome will die. The dead spores in the F1 and F2 tetrads (Pfliegler et al., 2012; Karanyicz et al., 2017) can be the products of lethal combinations of chromosome losses in the subgenomes. The loss of another chromosome from the subgenome from which the MAT-carrying chromosome was lost is not entailed with similar risk but can render additional chromosomes of the partner subgenome essential for viability. Thus, the spores that lose chromosomes from the same subgenome have better chances to remain viable. Because of the differences in gene contents even between chromosomes considered homeologous, random losses of chromosomes are likely to eliminate essential genes and thus lead to non-functional genomes. Retaining mainly the chromosomes of one subgenome and reducing preferentially the other subgenome also prevents eventual clashes of the different regulatory networks of the hybridized species (e.g., Tirosh et al., 2010; Schraiber et al., 2013; Metzger et al., 2017). It is pertinent to mention here that structural and regulatory incompatibilities have been assumed to shape the highly diverse alloaneuploid genomes in the S. pastorianus lager strains (for a review, see Krogerus et al., 2017a). The question remains as to why the same subgenome loses its MAT chromosomes in different hybrids of the same species pair. A similar model can be proposed for the postzygotic genome evolution during the vegetative propagation of the hybrid cells.

The combination of the gene pools of two or even more Saccharomyces species by natural mating of cells and the postzygotic chimerisation of the hybrid genomes by natural meiotic and mitotic segregation have great potentials in breeding of novel production strains for food industry. These processes generate a great diversity of new phenotypes without producing GMO strains with targeted genetic modifications. Both hybridisation and segregation are natural processes that take place in the nature without human intervention. The role of the human activity is to guide these processes in experimental conditions so that new phenotypes useful in fermentation and food industry can emerge.

The hybrids differ in phenotype from the parental species, but the differences can be both favorable and unfavorable. During the chimerisation process the nascent hybrid can get rid of “bad” genes or alleles, retain and combine their “good” counterparts and create new regulatory networks allowing better adaptation to technological environments. From the technological point of view, the allodiploid hybrids and the chimerised end products of the postzygotic genome evolution are better suited to the technological demands because their genomes are more stable than those of the intermediate forms (Figure 4). Examples of successful applications of these processes are shown in Tables 1–3, and certain major conclusions are discussed in the following sections.

The author confirms being the sole contributor of this work and has approved it for publication.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.