JOURNAL OF BACTERIOLOGY,0021-9193/99/$04.0010

Oct. 1999, p. 6488–6496 Vol. 181, No. 20

Copyright © 1999, American Society for Microbiology. All Rights Reserved.

Horizontal Transfer of Genetic Material among Saccharomyces YeastsGAELLE MARINONI,1 MARTINE MANUEL,1 RANDI FØNS PETERSEN,1 JEANNE HVIDTFELDT,1

PAVOL SULO,2 AND JURE PISKUR1*

Department of Microbiology, Technical University of Denmark, DK-2800 Lyngby, Denmark,1 andDepartment of Biochemistry, Comenius University, SK-842 15 Bratislava, Slovakia2

Received 25 May 1999/Accepted 11 August 1999

The genus Saccharomyces consists of several species divided into the sensu stricto and the sensu lato groups.The genomes of these species differ in the number and organization of nuclear chromosomes and in the sizeand organization of mitochondrial DNA (mtDNA). In the present experiments we examined whether theseyeasts can exchange DNA and thereby create novel combinations of genetic material. Several putative haploid,heterothallic yeast strains were isolated from different Saccharomyces species. All of these strains secreted ana- or a-like pheromone recognized by S. cerevisiae tester strains. When interspecific crosses were performed bymass mating between these strains, hybrid zygotes were often detected. In general, the less related the twoparental species were, the fewer hybrids they gave. For some crosses, viable hybrids could be obtained byselection on minimal medium and their nuclear chromosomes and mtDNA were examined. Often the frequencyof viable hybrids was very low. Sometimes putative hybrids could not be propagated at all. In the case of sensustricto yeasts, stable viable hybrids were obtained. These contained both parental sets of chromosomes butmtDNA from only one parent. In the case of sensu lato hybrids, during genetic stabilization one set of theparental chromosomes was partially or completely lost and the stable mtDNA originated from the same parentas the majority of the nuclear chromosomes. Apparently, the interspecific hybrid genome was genetically moreor less stable when the genetic material originated from phylogenetically relatively closely related parents; bothsets of nuclear genetic material could be transmitted and preserved in the progeny. In the case of moredistantly related parents, only one parental set, and perhaps some fragments of the other one, could be foundin genetically stabilized hybrid lines. The results obtained indicate that Saccharomyces yeasts have a potentialto exchange genetic material. If Saccharomyces isolates could mate freely in nature, horizontal transfer ofgenetic material could have occurred during the evolution of modern yeast species.

The genus Saccharomyces includes numerous species (2).These have been divided into sensu stricto and sensu latogroups (11). The sensu stricto group includes S. cerevisiae, S.bayanus, S. pastorianus, and S. paradoxus, and the sensu latogroup includes S. exiguus, S. castellii, S. unisporus, S. dairenen-sis, S. servazzii, and S. kluyveri (11). Yeasts belonging to thesensu stricto group are interfertile and represent sister species(15). Several natural hybrids, which are a product of interspe-cific crosses within the group, have been described (9, 12). Onthe other hand, sensu lato yeasts represent a far more heter-ogenous group of not well characterized yeasts (10). In gen-eral, a good definition of what a yeast species is still lacking.

The genome of S. cerevisiae has been studied intensively, andthe complete sequence has been determined. S. cerevisiae has16 chromosomes (for a review, see reference 4), and the mi-tochondrial DNA (mtDNA) molecule varies in size from 75 to85 kb (3a). A part of the S. cerevisiae genome represents anancient duplication (23). Other Saccharomyces sensu strictoyeasts exhibit a karyotype and genome size similar to those ofS. cerevisiae (22). The gene organization, including the geneorder, seems to be rather conserved among the sensu strictoyeasts. Therefore, the chromosomes are believed to be at leastpartially homologous (7, 20). mtDNA of sensu stricto yeastscontains many guanine-cytosine clusters, and the size is greaterthan 60 kb. S. paradoxus has a 75-kb mtDNA molecule, and S.pastorianus and S. bayanus have a 65-kb one (19). The coding

parts of the mitochondrial genes within the sensu stricto groupshow less than 5% sequence diversity (5).

Sensu lato yeasts generally contain a smaller number ofchromosomes (22). The two extremes are S. kluyveri, which has7 chromosomes, and S. exiguus, which is believed to contain 16(17). The four remaining sensu lato species lie between thesetwo extremes. S. dairenensis and S. castellii display 9 bands inpulsed-field gel electrophoresis analysis, and S. servazzii and S.unisporus display 12 (17). Sensu lato yeasts’ mtDNAs do notcontain an excess of guanine-cytosine clusters and are smallerthan 50 kb, i.e., 48 kb for S. dairenensis, 29 kb for S. servazziiand S. unisporus, 26 kb for S. castellii, and 23 kb for S. exiguus(19). The gene order varies a lot, and the coding regions of themitochondrial genes within the sensu lato group show 10 to20% sequence diversity (17a). Hence, each Saccharomyces spe-cies has a characteristic karyotype (17) and also a specificrestriction pattern of its mtDNA (19).

In principle, the definition of a yeast species should be basedon the concept of genetic isolation, as is the case for plants andanimals. This means that members of the same species areinterfertile, whereas genetically isolated species are not. How-ever, so far interfertility among Saccharomyces yeasts has notbeen examined in detail except among Saccharomyces sensustricto species. As mentioned before, sensu stricto yeasts canmate and generate viable hybrids. The best-described exampleof a yeast hybrid is the lager brewer’s yeast, S. pastorianus(synonym, S. carlsbergensis), which arose upon a mating eventbetween baker’s yeast, S. cerevisiae, and an unknown yeastbelonging to the S. bayanus complex. This hybrid is an allote-traploid displaying chromosomes from both parents, whereasits mtDNA molecule was inherited from the non-S. cerevisiaeparent (9, 19). However, two other native hybrids have been

* Corresponding author. Mailing address: Department of Microbi-ology, Technical University of Denmark, Building 301, DK-2800Lyngby, Denmark. Phone: (45) 45 252518. Fax: (45) 45 932809. E-mail:imjp@pop.dtu.dk.

6488

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

described recently also (12). Thus, two genomes can coexist inthe same cell, and it is likely that sensu stricto yeasts may beable to exchange their genetic material in nature. Several yearsago it was reported that S. cerevisiae and S. kluyveri couldrecognize each other’s pheromones and could generate hybridswith a decreased viability. However, these hybrids were notstudied in detail (1, 13). Because genetic tools, such as theavailability of haploid and auxotrophic strains, were poorlydeveloped for non-S. cerevisiae yeasts, other approaches wereused to define species among these yeasts. A major advance inour present understanding of yeast systematics has resultedfrom studies on DNA relatedness (10) and from nucleic acidsequence comparisons (8, 11). However, assessing phyloge-netic relationships by using sequence analyses of short piecesof DNA makes sense only if the tested yeasts are geneticallyisolated and of monophyletic origin.

Interspecific mating has previously been performed success-fully only between sensu stricto yeasts, and only the phenotypesof the resulting hybrids were analyzed (15). So far only limitedinformation on mating spores among sensu lato yeasts exists(16). In this study, we report that several Saccharomyces yeastscan mate at low frequency and produce viable hybrids. If sucha horizontal transfer of genetic material occurs in nature, mod-ern yeast species may contain DNA of polyphyletic origin.Thus, sequencing studies alone may sometimes be misleadingwhen used for the determination of phylogenetic relationshipsamong different yeasts.

MATERIALS AND METHODS

Strains. The yeast strains used in this project are listed in Table 1. Several areheterothallic and apparently also haploid. Some of the heterothallic non-S.cerevisiae strains were selected by testing isolates from various collections forpheromone secretion. Some homothallic strains were mutated by ethyl methane-sulfonate (EMS) mutagenesis or gene disruption to acquire heterothallic strains.

Media for routine growth. The yeast strains were grown at 25°C in the follow-ing media. YPD medium consisted of 1% yeast extract, 1% Bacto Peptone, and

2% glucose; minimal medium (SD) consisted of 1% succinic acid, 0.6% NaOH,0.67% yeast nitrogen base (Difco) without amino acids, and 2% glucose; sporu-lation medium consisted of 1% potassium acetate, 0.1% Bacto Yeast Extract,and 0.05% glucose. When necessary, these media were solidified with 2% agar.

EMS mutagenesis. The ho2 mutations and auxotrophic markers were intro-duced by EMS mutagenesis. One milliliter of stationary-phase yeast culture(either homothallic spores or prototrophic vegetative cells) was exposed to EMS(25 ml for Y188 and Y239, 30 ml for Y327 and Y328, and 35 ml in the case ofY323). A 200-ml portion of cells was transferred immediately to 8 ml of 5%sodium thiosulfate to quench the reaction (tube 1), but the rest of cells wereincubated at 30°C for 45 min and 200 ml of cells was transferred to tubes 2 and3 containing 8 ml of 5% thiosulfate. Finally, the rest of the cells were exposed toEMS in the same manner for 2 h (tubes 4 and 5).

Cells from tubes 1, 2, and 4 were plated on YPD medium to determine thepercentages of surviving cells. The chemical treatment was considered efficientwhen survival was lower than 50%. Tubes 3 and 5 were centrifuged (3,000 3 g for4 min), and the cells were washed twice with sterile water. Thereafter, the cellswere resuspended in 1 ml of YPD medium and incubated at 25°C for 60 min. Thesuspension was spread onto YPD plates (ca. 200 cells/plate), and after 3 days ofincubation, the plates were replicated onto either sporulation medium (whenscreening for ho2 mutants) or SD (when screening for auxotrophs). Putativeauxotrophs were further examined to identify the induced auxotrophic markers.

Halo assay. When pheromones from yeast cells of the opposite mating type arepresent in the medium, haploid heterothallic yeast cells undergo G1 phase arrest.Such cells do not divide further. To test for pheromone secretion, approximately105 cells of a haploid S. cerevisiae tester strain (MT502, MT503, MT504, orMT505) were spread as a lawn on a YPD plate. MT502 and MT505 were usedas a tester for a factor secretion, whereas MT503 and MT504 were used as atester for a factor secretion. Then various yeast colonies were transferred aspatches to such plates. Clear zones around patches after overnight incubationindicated inhibition of growth of the tester strain.

Induction of heterothallism. The diploid S. castellii strain Y188 was sporu-lated, and separation of spores was achieved in the following way. The asci wereresuspended in 200 ml of Zymolyase 100T at a concentration of 0.2 mg/ml andincubated while shaking for 30 min at 25°C. Approximately 100 ml of sterile glassbeads was then added, and the tube was incubated for another 30 min. Afteraddition of 0.2 ml of 0.1% Triton X-100, the tube was vortexed for 1 min. Sporeseparation was checked under a microscope before spores were washed twicewith sterile sodium phosphate buffer, pH 7.0. Thereafter the spores were mu-tagenized with 25 ml of EMS by the procedure described above.

The treated cells were grown on YPD medium and replicated onto sporulationmedium, and after several days they were checked for sporulation under UVlight. In Saccharomyces yeasts, a sporulated culture exhibits fluorescence when

TABLE 1. Strains used in this study

Strain numbera Species or strain Mating type Genotype or description Reference or sourcec

Y169 S. cerevisiae MATa lys1 This laboratoryY170 S. cerevisiae MATa lys1 This laboratoryY184 S. cerevisiae MATa ade1 his1 This laboratoryY185 S. cerevisiae MATa ade1 his1 This laboratoryY266 S. cerevisiae MATa lys1 r2 Y170Y267 S. cerevisiae MATa lys1 r2 Y170Y166 S. bayanus Diploid Homothallic, ura3-1 MCYC 623-6C (R. Borts)Y244 S. bayanus MATa ura32 G418R Y166 transformed with P161Y245 S. bayanus MATa ura32 G418R Y166 transformed with P161Y327 S. exiguus MATa Spontaneous ho2 CBS 2141Y328 S. exiguus MATa Spontaneous ho2 CBS 2141Y339 S. exiguus MATa his2 Y327Y345 S. exiguus MATa arg2 Y328Y188 S. castellii Diploid Homothallic CBS 4310Y239 S. castellii MATa Induced ho2 Y188Y257 S. castellii MATa met2 Y239Y323 CBS 6463b MATa Spontaneous ho2, r2 CBS 6463Y344 CBS 6463 MATa ura32 r2 Y323Y280 Hybrid Hybrid Prototroph Y244 3 Y169Y500 Hybrid Hybrid Prototroph Y257 3 Y245Y507 Hybrid Hybrid Prototroph, r2 Y184 3 Y344MT502 S. cerevisiae MATa sst2-1 met1 leu2-3,112 his3 can1 M. T. HansenMT503 S. cerevisiae MATa sst2-1 met1 leu2-3,112 his3 can1 M. T. HansenMT504 S. cerevisiae MATa leu2-3,112 gal2 sst1 V. McKayMT505 S. cerevisiae MATa leu2-3,112 gal2 sst1 V. McKay

a Y, J. Piskur collection; MT, T. Nilsson-Tillgren collection.b CBS 6463 (Y323) was previously classified as S. dairenensis, but it actually represents a novel species (17b).c MCYC, Microbiology Collection of Yeast Cultures, Madrid, Spain; CBS, Centraalbureau voor Schimmelculture, Delft, Holland.

VOL. 181, 1999 HORIZONTAL GENE TRANSFER AMONG YEASTS 6489

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

excited by UV light at 305 nm. Colonies which were not fluorescent were checkedfor sporulation by microscopy. Colonies which did not give rise to spores wereassumed to be either haploids mutated in the HO gene or diploids mutated in agene necessary for the sporulation process. In the next step such mutants werechecked for secretion of pheromones by using the S. cerevisiae tester strains. Ahalo around a colony was taken as evidence that the strain exhibited sexualactivity and presumably was heterothallic and haploid.

Disruption of the HO gene. The homothallic, diploid S. bayanus strain Y166was transformed with the plasmid P161, which carries a part of the S. cerevisiaeHO gene. The plasmid was constructed in the following way. The initial vectorwas the pCH217 plasmid (from Chris Harfield, University of Leicester, UnitedKingdom), which contains a 2-kb insert of the ATP1 gene conferring resistanceto geneticin, G418R, at a concentration of 150 mg/liter (see also reference 18).The ATP1 gene is flanked on both sides by several unique restriction sites, e.g.,PstI and BamHI, which both map 39 from the ATP1 gene. A 0.75-kb PstI-BglIIfragment of the S. cerevisiae HO gene originating from a plasmid carrying thecomplete HO gene (14) was inserted into pCH217, which had previously beendigested with PstI and BamHI. The newly constructed plasmid was named P161.The 0.75-kb HO fragment carries BamHI and Eco47III restriction sites, whichare present as unique sites in P161. For transformation purposes P161 was cutwith BamHI or Eco47III and introduced into a culture containing spores of Y166by the usual yeast transformation procedure. The plasmid P161 was expected tobe inserted into the HO gene and generate two defective copies of this gene.Colonies resistant to geneticin were examined for the structure of the HO regionby Southern analysis and for secretion of pheromones. Diploid transformed cellswere sporulated, and haploid lines were obtained after dissection of tetrads.

Mass mating test. The auxotrophic parental strains were grown independentlyin liquid YPD medium overnight. Stationary-phase a and a cells of differentmutants, corresponding to approximately 0.5 3 108 cells, were mixed togetherand then dropped onto YPD plates. After several hours of incubation at 25°C,the yeasts were checked by microscopy for the formation of zygotes. Whenzygotes were detected, approximately 107 cells were taken from the YPD plateand plated onto SD, or SD plus additives, in order to screen for viable hybrids.

Selection procedure. When both parents were auxotrophs, viable hybrids wereselected on the selective SD medium. Cells were grown for 3 to 5 days at 25°C.The parental strains were also incubated on SD medium in order to check forrevertants.

DAPI staining for nucleus detection. Suspensions of cells containing zygoteswere stained with DAPI (49,6-diamidino-2-phenylindole), which causes DNA tofluoresce bluish-white. Approximately 105 cells were added to 20 ml of DAPIsolution (1 mg of DAPI per 1 ml of H2O) and boiled for 5 min. Under afluorescence microscope it was possible to determine whether the two parentalnuclei in zygotes had fused. This method indicated whether zygotes were homo-or heterokaryons.

Analysis of nuclear chromosomes. Chromosomes of various putative hybridswere isolated and separated by pulsed-field gel electrophoresis (contour-clampedhomogeneous electric field [CHEF] gel electrophoresis). The usual conditionsfor preparation of chromosomal DNA from the Saccharomyces yeasts werefollowed (17). A CHEF-DR II apparatus (Bio-Rad, Richmond, Calif.) was usedfor separation of chromosomes. The electrophoresis gels had an agarose con-centration of 1%, and the electrophoresis buffer was 0.53 TBE (103 TBE is 1.0M Tris, 0.9 M boric acid, and 0.01 M EDTA) cooled to 14°C. Electrophoresis wascarried out at 150 V for 8 h with a switching time of 240 s and then for 10 h witha switching time of 160 s followed by 14 h with a switching time of 90 s and finally6 h with a switching time of 60 s. After electrophoresis, the gels were stained withethidium bromide for visualization of chromosomal DNA.

Analysis of mtDNA. mtDNA was isolated according to a fast procedure de-scribed by Defontaine et al. (3). In some cases, mtDNA was isolated by the CsClmethod (19). mtDNA was cut with restriction enzymes, and fragments wereseparated on a gel. Mitochondrial fragments were visualized with ethidiumbromide, or the gel was blotted onto a membrane and the mtDNA bands werevisualized by hybridization with a labelled probe based on CsCl-purified mtDNA.

RESULTS

Construction of haploid heterothallic strains. Two yeaststrains belonging to the same Saccharomyces species can becrossed successfully if they are of opposite mating types. Across can occur between homothallic spores or by crossingheterothallic strains. In the present study it was possible toobtain, isolate, or construct strains belonging to the sensu latoyeasts that were maters and that carried recessive auxotrophicmutations. The strains were tested for secretion of phero-mones on the Saccharomyces tester strains. All strains used inthis study could arrest S. cerevisiae in the G1 phase (Table 2).These strains were used in mass mating experiments betweendifferent species.

S. castellii heterothallic strains were isolated after EMS mu-

tagenesis on the Y188 diploid strain as described above. Y239secreted only one type of pheromone (Table 2). Furthermore,a recessive mutation was induced in this strain, and thereforethis strain probably is a haploid. The resulting methionine-requiring strain, Y257, was used in mass mating experiments.

S. bayanus heterothallic strains were isolated after disrup-tion of the HO gene in the Y166 strain. The P161 plasmid,which was cut with Eco47III, recombined into the HO gene insuch a way as to generate two mutant copies of the HO gene(data not shown). The crossovers most likely took place be-tween two regions originating from two different species, i.e., S.cerevisiae and S. bayanus. Apparently, the homology was suf-ficiently high to promote homologous recombination. Whentransformed Y166 colonies sporulated, haploid strains, such asY244 and Y245, each secreting only one type of pheromone,were obtained upon tetrad dissection (Table 2). Transforma-tion of sensu lato strains with P161 was not successful, possiblybecause homology between the S. cerevisiae sequences and thetarget sequences was too low.

S. exiguus strains Y339 and Y345, as well as CBS 6463 (strainY344), were obtained from different yeast collections as natu-ral isolates of heterothallic strains. Two of these three strainssecreted one type of pheromone, but Y339 secreted stronglya-like, and weakly a-like, pheromone (Table 2). Recessivemutations were induced successfully in these strains, and thusthey also are likely to be haploid.

The strains described above were then tested for geneticstability of the induced mutations. All strains listed in Table 1gave a very low frequency of revertants, i.e., less than 1027, andwere therefore suitable for mating experiments. Table 2 showsthat all heterothallic strains secreted sexual pheromones thatcould be recognized by the hypersensitive S. cerevisiae testerstrains. Thus, the action of these pheromones is not restrictedto only one species.

Mating. The strains described above were then tested fortheir ability to undergo further steps in the mating process, i.e.,abilities to fuse and produce zygotes. Various crosses wereperformed, and the presence of different cell types, especiallyzygotes, was checked by microscopy (Fig. 1 and Table 3). Zy-gotes appeared at various frequencies after strains were ex-posed to one another for 5 to 8 h.

Within the sensu stricto group, mating occurred at highfrequencies (ca. 22 to 40% of cells in the mating mix werepresent as zygotes). Intra- and interspecific crosses gave asimilar frequency of zygotes (Table 3, crosses A and C). Ho-mocrosses, i.e., mating between yeasts having the same matingresponse on the tester strains, did not result in any zygotes. Inshort, the mating between S. cerevisiae and S. bayanus was very

TABLE 2. Secretion of pheromones among Saccharomyces strains

Strain Species orstrain

Pheromone secretiona on:Mating

typeMT502(MATa)

MT503(MATa)

MT504(MATa)

MT505(MATa)

Y184 S. cerevisiae 111 X X 1 MATaY185 S. cerevisiae X 111 1 X MATaY169 S. cerevisiae X 111 1 X MATaY170 S. cerevisiae 111 X X 1 MATaY244 S. bayanus 111 X X 1 MATaY245 S. bayanus X 111 1 X MATaY339 S. exiguus 111 1 X 1 MATaY345 S. exiguus X 111 X X MATaY257 S. castellii X 111 1 X MATaY344 CBS 6463 X 111 X X MATa

a Symbols: X, no response; 1, weak signal; 111, strong signal.

6490 MARINONI ET AL. J. BACTERIOL.

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

FIG. 1. Microscopic observation of interspecific zygotes obtained after a cross between S. cerevisiae (Y184) and S. exiguus (Y345). (A) Parental strain Y184. (B)Parental strain Y345. Note that no cells in the mixture have a zygote-like shape. (C) A mixture in which the two parental strains had been in contact for 6 h. Bothparental cells and cells with a zygote-like shape are visible.

VOL. 181, 1999 HORIZONTAL GENE TRANSFER AMONG YEASTS 6491

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

efficient and mating type dependent. S. cerevisiae was alsosuccessfully crossed with all sensu lato isolates. However, thefrequency of zygotes appeared to be lower than that for crosseswithin the sensu stricto group. Most of the time, the appear-ance of zygotes was mating type dependent but some ho-mocrosses were also observed, e.g., Y185 3 Y257 and Y185 3Y344 (Table 3, cross D).

In the case of sensu lato yeasts, zygotes were obtained onlyfrom homocrosses. In the case of crosses between yeasts havingthe opposite mating type, such as Y339 3 Y257 or Y339 3Y344, no zygotes were observed. In the sensu lato group, theinterspecific crosses were apparently not mating type depen-dent. Note that these yeasts were tested several times on S.cerevisiae tester strains (Table 2), and they secreted only onetype of pheromones as recognized by S. cerevisiae. The hybridnature of some putative hybrids resulting from homocrosseswas confirmed by chromosome analysis (see later sections).Similar homocrosses were observed and reported previouslyamong wild-type yeast isolates (21). In general, frequencies ofzygotes were lower in the case of interspecific crosses amongsensu lato yeasts than in other crosses. However, these differ-ences between zygote frequencies are difficult to interpret,because the strains used are perhaps not perfectly heterothal-lic, i.e., may have a leaky ho2 allele, or they may secretedifferent amounts of pheromones. Lower frequencies may be aconsequence of variations in the structure of pheromones be-longing to different species. These experiments clearly demon-strated that heterothallic yeasts belonging to the genus Sac-

charomyces can recognize each other and that the initial stepsof mating can take place among different species.

Karyogamy. Zygotes from a few interspecific crosses wereexamined to ascertain the number of nuclei per cell. Afterpaired parental strains were incubated for 5 to 8 h the cellswere DAPI stained and examined by fluorescence microscopy.If they contained one nucleus they were designated homokary-ons, and if two parental nuclei were present in each cell theywere designated heterokaryons. After 6 h the zygotes pos-sessed only one nucleus, and no heterokaryons were found(Table 3). Therefore, fusion of nuclei occurred almost imme-diately upon generation of interspecific hybrid zygotes.

Frequency of viable hybrids. In the following experiment,various yeasts were crossed and selection for viable hybrids wasundertaken. Crosses were performed between strains of differ-ent species carrying auxotrophic markers and cells plated onselective medium. In most cases, the mixture contained a sub-stantial proportion of zygotes (see also Table 3). Note thatwhen a hybrid colony appeared on the selective medium, theputative hybrid had already undergone a large number of di-visions. Results showing a number of putative hybrids are pre-sented in Table 4. Some crosses, such as Y184 3 Y339, Y339 3Y257, Y339 3 Y344, Y184 3 Y257, and Y184 3 Y345, whichprovided only a single viable hybrid or no hybrids, are notshown in Table 4. However, later on some of these hybrids,which appeared at a very low frequency, were also analyzed forthe structure of their genomes. In all cases, the frequency ofviable hybrids was far lower than the frequency of zygotesobserved under the microscope. In the case of an intraspecificcross between two S. cerevisiae strains, most zygotes seem tohave given viable diploids (Table 4). However, even in the caseof crosses between S. cerevisiae and S. bayanus, which arerelatively closely related, only ca. one-tenth of the observedzygotes seem to have given viable hybrid colonies. The fre-quency of viable hybrids was several orders of magnitude lower

TABLE 4. Frequency of viable interspecific hybrids

Crossa No. of putativehybrids on SDb

Frequency of putativeviable hybridd

AY185 3 Y170 ;104 32 3 1022

Y184 3 Y245 1,000 3.8 3 1022

Y185 3 Y244 1,500 4.5 3 1022

Y169 3 Y244 1,500 4.5 3 1022

Y170 3 Y245 1,000 3.3 3 1022

BY185 3 Y339 6c 6.7 3 1026

Y184 3 Y344 50c 1024

Y185 3 Y344 70c 4.7 3 1025

CY345 3 Y257 ;103 ;3.3 3 1023

Y345 3 Y344 ;103 ;6.7 3 1024

Y257 3 Y344 5 1 50e 7.8 3 1025

a A, crosses within the sensu stricto group; B, crosses between yeasts belongingto the sensu stricto and sensu lato groups; C, crosses between yeasts belonging tothe sensu lato group. Note that only interspecific crosses, which gave viablehybrids, are shown in this table.

b All of the parental strains were tested for revertants on SD, and the reversionrate was shown to be lower than 1027 for each of them. Therefore, when a colonyappeared on SD, it was likely to be a hybrid because only approximately 107

parental cells, including zygotes, were plated on the selective plate.c Small colonies.d Frequencies of putative viable hybrids are the ratio between the number of

colonies on SD plates and the presumed number of hybrids plated out.e Five large and 50 small colonies.

TABLE 3. Appearance of zygotes in intra- and interspecies crosses

Crossa Mating typeFrequency

ofzygotesb

Presence ofhomo- or

heterokaryonsc

AY184 3 Y185 MATa 3 MATa 0.40 n.p.Y169 3 Y170 MATa 3 MATa 0.30 n.p.Y184 3 Y169 MATa 3 MATa 0.22 n.p.Y244 3 Y245 MATa 3 MATa 0.30 n.p.

B (Y339 3 Y345) MATa 3 MATa 0.05 n.p.

CY184 3 Y245 MATa 3 MATa 0.26 n.p.Y185 3 Y244 MATa 3 MATa 0.33 n.p.Y169 3 Y244 MATa 3 MATa 0.33 n.p.Y170 3 Y245 MATa 3 MATa 0.30 n.p.

DY184 3 Y345 MATa 3 MATa 0.08 HomokaryonsY184 3 Y257 MATa 3 MATa 0.10 HomokaryonsY184 3 Y344 MATa 3 MATa 0.05 HomokaryonsY185 3 Y339 MATa 3 MATa 0.09 n.p.Y185 3 Y257 MATa 3 MATa 0.19 n.p.Y185 3 Y344 MATa 3 MATa 0.15 n.p.

EY345 3 Y257 MATa 3 MATa 0.03 HomokaryonsY345 3 Y344 MATa 3 MATa 0.15 HomokaryonsY257 3 Y344 MATa 3 MATa 0.07 Homokaryons

a A, intraspecific crosses within the sensu stricto group; B, intraspecific crosseswithin the sensu lato group; C, interspecific crosses between yeasts belonging tothe sensu stricto group; D, interspecific crosses between yeasts belonging to thesensu stricto and sensu lato groups; E, interspecific crosses between yeastsbelonging to the sensu lato group.

b Note that the frequency of zygotes is estimated on approximately 1,000 cells.c n.p., not performed.

6492 MARINONI ET AL. J. BACTERIOL.

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

in the case of crosses employing sensu lato yeasts. Note thatthese sensu lato yeasts are not very closely related to eachother or to S. cerevisiae (11). Apparently, a majority of zygoteswhich appeared during the first hours after mating were notstable and did not give rise to viable prototrophic progeny.

Analysis of the hybrid progeny. Some of the putative hybridswhich formed colonies on minimal medium were analyzed forthe organization of their genomes. Chromosomal DNA wasextracted and separated by pulsed-field gel electrophoresis,and the patterns were compared to those of the parentalstrains. Also mtDNA molecules were purified and subjected torestriction analysis. The patterns obtained were compared tothose of the parental strains. Results are shown in Tables 5 and6 and in Fig. 2. Hybrids obtained from the crosses between S.cerevisiae and S. bayanus always contained both parental chro-mosomes (Tables 5 and 6 and Fig. 2). Only occasionally was asingle chromosome band belonging to one parental set missingor did it exhibit a changed size (data not shown). Apparently,the hybrid situation was genetically stable and both parentalchromosomes could cohabit within the same nucleus. How-ever, if both parental strains contained mtDNA, then only theS. cerevisiae mtDNA molecule was detected among the prog-eny (Table 6). Only when the S. cerevisiae parent was a petitemutant, and thus did not contain any functional mtDNA, wasthe S. bayanus mtDNA molecule transmitted to the progeny(Table 6). In principle a hybrid could contain, propagate, andexpress either of the two parental mtDNA molecules.

In hybrids for which at least one of the parents belonged tothe sensu lato group, the situation regarding the viability andgenome structure was different from that in the previous case.

Most hybrid colonies which appeared on selective medium hadlow viability when transferred to fresh selective medium. Inaddition, many of these hybrids showed poor growth even onrich medium. For example, hybrids originating from the crossbetween S. cerevisiae and CBS 6463 were initially representedby small colonies, which were subsequently difficult to propa-gate even on YPD medium. Only some of the hybrids could bepropagated sufficiently to allow analyses of their nuclear andmitochondrial genomes. In general, only one parental set ofchromosomes was found in hybrids of sensu stricto and sensulato yeast species (Table 5). This chromosomal set was alwaysaccompanied by the mtDNA molecule having the same paren-

TABLE 5. Analysis of the hybrid progeny

Cross Hybrid no. Nuclear pattern Mitochondrialpattern

Growth ona:

YPD medium SD

Y184 3 Y245 1–10 S. cerevisiae and S. bayanus S. cerevisiae 111 111Y169 3 Y244 1–10 S. cerevisiae and S. bayanus S. cerevisiae 111 111Y185 3 Y339 1–6 S. exiguus S. exiguus 1 1Y184 3 Y344 1–20 CBS 6463 r2 1 1/2b

Y345 3 Y257 1–10 S. exiguus S. exiguus 111 111–20 S. castellii S. castellii 1 1

Y345 3 Y344 1–10 S. exiguus S. exiguus 1 111–20 CBS 6463 r2 1 1

Y184 3 Y344 Y507 CBS 6463 plus one bandfrom S. cerevisiae

r2 1 1/2

Y257 3 Y245 Y500 Mixed parental and newbands

S. bayanus 111 111

a 111, very good growth; 2, no growth; 1/2, very weak growth; 1, weak growth.b Most of these hybrids had reduced viability on any medium, and they were difficult to propagate.

TABLE 6. Sensu stricto crosses

Cross Characteristics Nuclearpatterna

Mitochondrialpatterna

Y244 3 Y169 S. cerevisiae r1 3S. bayanus r1

Both parentalchromosomes

S. cerevisiae

Y245 3 Y170 S. cerevisiae r1 3S. bayanus r1

Both parentalchromosomes

S. cerevisiae

Y266 3 Y245 S. cerevisiae r2 3S. bayanus r1

Both parentalchromosomes

S. bayanus

Y267 3 Y245 S. cerevisiae r2 3S. bayanus r1

Both parentalchromosomes

S. bayanus

a Over 100 hybrids from each cross were checked for their nuclear and mito-chondrial genome.

FIG. 2. Karyotypes of two interspecific hybrids and their parental strains.Lanes: 1, S. bayanus; 2, S. cerevisiae; 3, hybrid between S. bayanus and S.cerevisiae, Y280; 4, S. castellii; 5, S. bayanus; 6, hybrid between S. bayanus and S.castellii, Y500. The interspecific hybrid produced by crossing yeasts belonging tothe sensu stricto group shown in lane 3 displays both sets of parental chromo-somes, whereas the interspecific hybrid produced by crossing a sensu stricto yeastand a sensu lato yeast shown in lane 6 displays most of the chromosomes fromthe sensu stricto parent, an extra chromosome band originating from the sensulato parent, and some bands of a modified size or an unknown origin (arrows).

VOL. 181, 1999 HORIZONTAL GENE TRANSFER AMONG YEASTS 6493

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

tal origin as the chromosomes (Table 5). It is interesting topoint out that in some crosses, such as Y345 3 Y257 orY257 3 Y344, the resulting hybrids had one set of chromo-somes, but originating from either one parent or the other.Only rarely were additional chromosome bands, in addition toone parental set, observed in some hybrids, such as Y500 andY507. Apparently such chromosomal bands were inheritedfrom the other parent, or their origin could not be deduced(Fig. 2). Sensu stricto-to-sensu stricto hybrids could sporulateat a very low frequency, and a majority of spores were notviable. A majority of sensu stricto-to-sensu lato and sensulato-to-sensu lato hybrids could not sporulate; for example,Y500 and Y507 did not sporulate.

DISCUSSION

The genus Saccharomyces contains a number of specieswhich differ from one another in various phenotypic charac-teristics (2). Among the most polymorphic characteristics arethe number, size, and organization of nuclear chromosomesand the size and organization of mtDNA (17, 19, 22). DNArelatedness studies and gene sequencing provided the first av-enues for understanding the evolutionarary relationshipsamong modern Saccharomyces yeasts (8, 11). So far, sequencecomparisons have been performed on genes coding for rRNAand because of technical reasons not on other genes (for areview, see reference 10). So far our understanding of evolu-tionary relationships has been based on the assumption thateach of the modern species is monophyletic and genetically

isolated from other species. If so, then the sequence of anypiece of the genome should reflect the evolutionary history ofthe whole genome, as well as a phylogenetic relationship withother species. However, if horizontal transfer of genetic mate-rial still operates among species, then analysis of a single geneand comparison of it with other genes may lead to misleadinginformation about evolutionary relationships.

The present project has established a laboratory model forstudying horizontal gene transfer among Saccharomyces yeasts.For this purpose, different mating-competent yeasts were pre-pared. Some of these strains, such as Y339, Y345, and Y344,were isolated upon screening natural isolates from differentcollections. Some others, such as Y244, Y245, and Y257, wereconstructed by gene disruption or mutagenesis.

S. cerevisiae strains sensitive to pheromones were able torecognize sexual factors secreted by any of the putative hap-loid, heterothallic strains of the tested Saccharomyces yeastspecies (Table 2). When these yeasts were exposed to eachother in many combinations, interspecific zygotes were de-tected (Fig. 1 and Table 3). Therefore, no absolute, interspe-cific barriers exist in the mating process among the Saccharo-myces yeasts. Apparently, different species could recognizeeach others’ pheromones, cells fused, and karyogamy tookplace (Tables 2 and 3). Interspecific hybrids were generatedand existed for at least a short time. While mating betweencells of yeasts from different species was possible, it happenedwith a lower frequency than intraspecific mating between cellsof strains of the same species, e.g., S. cerevisiae (Table 3). Ingeneral, the less related the species were, the lower the fre-

FIG. 3. Schematic illustration of mating events within the genus Saccharomyces. (A) A hybrid produced by crossing two sensu stricto yeasts contains chromosomesfrom both parents but mtDNA from only one parent. In the case of a sensu stricto-to-sensu lato cross (B) or a sensu lato-to-sensu lato cross (C), predominantly onlyone parental set, plus some fragments from the other parent, are observed.

6494 MARINONI ET AL. J. BACTERIOL.

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

quency of zygotes was (Table 3). The S. cerevisiae mating typesystem was followed in crosses where at least one sensu strictoyeast was involved, i.e., MATa cells from one species matedwith MATa cells from another, and vice versa (Table 3). In anumber of crosses, particularly when sensu lato species werecrossed, cells classified as MATa of one species mated withMATa cells from another species, or MATa cells mated withMATa cells (Table 3). This observation is difficult to explain.However, bisexual behavior of Saccharomyces yeasts has beenreported earlier (21). Nevertheless, the most interesting resultof these crosses is that hybrids were generated at all.

Saccharomyces sensu stricto yeasts are believed to have ho-mologous chromosomes; i.e., the order of genes is largely pre-served among different species (7, 20). Interestingly, viablehybrids between two sensu stricto yeasts, S. cerevisiae and S.bayanus, displayed both sets of parental chromosomes (Tables5 and 6 and Fig. 2). These hybrids were stable and could ingeneral propagate themselves through mitosis during manygenerations without undergoing any apparent rearrangementsof their nuclear genome. It seems that both parental sets ofchromosomes were compatible and could coexist in the samecell (Fig. 3). On the other hand, mtDNA was inherited fromonly one parent. If both parental mtDNA molecules werepresent in the zygotes, the S. cerevisiae mtDNA moleculesoutcompeted the S. bayanus mtDNA molecule and were pref-erentially transmitted to the progeny. When the S. cerevisiaeparent did not have a functional mtDNA molecule, the S.bayanus mtDNA molecule was transmitted successfully andpersisted among the progeny (Table 6). Thus, the hybrid nu-clear background can accommodate either of the parentalmtDNA molecules. However, the frequency of viable hybrids,i.e., the frequency of zygotes able to go through mitosis andgenerate a yeast colony, was lower in the case of interspecificcrosses than in the case of intraspecific crosses (Table 4). Agreat part of zygotes obtained upon interspecific mating didnot give viable hybrids. Apparently, a weak species barrierexists between S. cerevisiae and S. bayanus. In any case, naturalhybrids of sensu stricto yeasts exist in nature (6, 12), andinterspecific matings may have contributed to the polymor-phism observed among sensu stricto yeasts.

Members of the sensu lato group have from 7 to 16 chro-mosomes, which so far are only poorly characterized. Preser-vation of gene order seems to be limited (17b). Furthermore,mtDNAs vary in size, in gene order, and also in their introns(17a, 19). In interspecific crosses, S. cerevisiae could give zy-gotes upon mating with several other yeast strains belonging tothe sensu lato group. However, interspecific zygotes producedby mating between yeasts belonging to the sensu stricto and thesensu lato groups appeared at a frequency several times lowerthan in the case of crosses within the sensu stricto group (Table3). Only a small fraction of these zygotes developed furtherand gave rise to viable hybrids. A similar situation existed inthe sensu lato-to-sensu lato crosses (Table 4). In general, pu-tative hybrids inherited nuclear DNA and mtDNA from onlyone of the two parents. In some cases, putative hybrids con-tained a complete set of chromosomes from one parent, as wellas extra chromosomes from the other parent (Fig. 2 and 3).Moreover, sometimes novel chromosome bands appeared.These results indicate that both parental sets of chromosomeswere not compatible within one cell; subsequently, during earlymitotic divisions, one set of chromosomes was wholly or par-tially lost (Fig. 3). Apparently, at most a limited number ofgenes from one of the parents was retained as a separatechromosome or perhaps by integration into one or more of thechromosomes of the other parent (Fig. 3).

In conclusion, interspecific hybrids could be obtained by

crossing yeasts belonging to the genus Saccharomyces. How-ever, genome integrity and stability seem to differ in hybridsproduced by crossing closely and less closely related yeasts. Inthe case of a close phylogenetic relationship, i.e., within thesensu stricto group, interspecific zygotes can be obtained at ahigh frequency, and some of them can give rise to stable hy-brids possessing both sets of parental chromosomes. Thus,such parental genomes can coexist. In the case of lesser phy-logenetic and structural relationships, as in crosses betweensensu stricto and sensu lato yeasts or crosses between differentsensu lato yeasts, interspecific zygotes appeared with low fre-quency, and only a very small fraction of them resulted ininterspecific hybrids. In these cases, the two parental sets ofchromosomes seem not to be compatible, and extensive rear-rangements occur. It appears that similar organization betweentwo parental sets of chromosomes correlates with the coexist-ence of two parental chromosome sets in the nuclei of hybridcells and their stability.

The results described above prove that horizontal transfer ofgenetic material is possible among the modern species of thegenus Saccharomyces. If this mechanism operated also in thepast, then the modern genomes may not have a simple mono-phyletic origin. Thus, when sequence analysis is used to deter-mine the relationship among Saccharomyces yeasts, severalindependent genes should be studied.

ACKNOWLEDGMENTS

This work was supported by grants from the Danish Natural ScienceResearch Council (SNF), the Carlsberg Foundation, the PlasmidFoundation, and the Novo Nordisk Foundation. Gaelle Marinoni andMartine Manuel acknowledge training support from the EU Erasmus/Socrates program and the Technical University of Denmark.

Judita Gartner is acknowledged for help in isolation of heterothallicmutants. Torsten Nilsson-Tillgren, Albert Kahn, and Gennadi Naumovare acknowledged for their comments on this work.

REFERENCES

1. Barker, E. R., and M. W. Miller. 1969. Some properties of Saccharomyceskluyveri. Antonie van Leeuwenhoek 35:159–171.

2. Barnett, J. A. 1992. The taxonomy of the genus Saccharomyces Meyen exReess: a short review for non-taxonomists. Yeasts 8:1–23.

3. Defontaine, A., F. M. Lecocq, and J. N. Hallet. 1991. A rapid miniprepmethod for the preparation of yeast mitochondrial DNA. Nucleic Acids Res.19:185.

3a.de Zamaroczy, M., and G. Bernardi. 1985. Sequence organization of themitochondrial genome of yeast—a review. Gene 37:1–17.

4. Goffeau, A., B. G. Barrell, H. Bussey, R. W. Davis, B. Dujon, H. Feldmann,F. Galibert, J. D. Hoheisel, C. Jacq, M. Johnston, E. J. Louis, H. W. Mewes,Y. Murakami, P. Philippsen, H. Tettelin, and S. G. Oliver. 1996. Life with6000 genes. Science 274:546–567.

5. Groth, C. 1998. Saccharomyces sensu stricto yeasts: characterization of mi-tochondrial DNA. Masters Thesis. University of Copenhagen, Denmark.

6. Groth, C., J. Hansen, and J. Piskur. A natural chimeric yeast containinggenetic material from three species. Int. J. Syst. Bacteriol., in press.

7. Hunter, N., S. R. Chambers, E. J. Louis, and R. H. Borts. 1996. The mis-match repair system contributes to meiotic sterility in an interspecific yeasthybrid. EMBO J. 15:1726–1733.

8. James, S. A., J. Cai, I. N. Roberts, and M. D. Collins. 1997. A phylogeneticanalysis of the genus Saccharomyces based on 18S rRNA gene sequences:description of Saccharomyces kunashirensis sp. nov. and Saccharomyces mar-tinae sp. nov. Int. J. Syst. Bacteriol. 47:453–460.

9. Kielland-Brandt, M., T. Nilsson-Tillgren, C. Gjermansen, S. Holmberg, andM. B. Pedersen. 1995. Genetics of brewing yeasts, p. 223–254. In A. H. Rose,E. Wheals, and J. S. Harrison (ed.), The yeasts, vol. 6. Academic Press,London, United Kingdom.

10. Kurtzman, C. P., and J. W. Fell. 1998. The yeasts: a taxonomic study, 4th ed.Elsevier, Amsterdam, The Netherlands.

11. Kurtzman, C. P., and C. J. Robnett. 1991. Phylogenetic relationships amongspecies of Saccharomyces, Debaryomyces and Schwanniomyces determinedfrom partial ribosomal RNA sequences. Yeast 7:61–72.

12. Masneuf, I., J. Hansen, C. Groth, J. Piskur, and D. Dubourdieu. 1998. Newhybrids between Saccharomyces sensu stricto yeast species found among wineand cider production strains. Appl. Environ. Microbiol. 64:3887–3892.

VOL. 181, 1999 HORIZONTAL GENE TRANSFER AMONG YEASTS 6495

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

13. McCullough, J., and I. Herskowitz. 1979. Mating pheromones of Saccharo-myces kluyveri: pheromone interactions between Saccharomyces kluyveri andSaccharomyces cerevisiae. J. Bacteriol. 138:146–154.

14. Meiron, H., E. Nahon, and D. Raveh. 1995. Identification of the heterothallicmutation in HO-endonuclease of S. cerevisiae using HO/ho chimeric genes.Curr. Genet. 28:367–373.

15. Naumov, G. I. 1996. Genetic identification of biological species in the Sac-charomyces sensu stricto complex. J. Ind. Microbiol. 17:295–302.

16. Naumov, G. I., E. S. Naumova, G. Marinoni, and J. Piskur. 1998. Geneticanalysis of S. castellii, S. exiguus and S. martiniae yeasts. Genetika 34:565–568.

17. Petersen, R. F., T. Nilsson-Tillgren, and J. Piskur. Karyotypes of Saccharo-myces sensu lato species. Int. J. Syst. Bacteriol., in press.

17a.Petersen, R. F. 1998. Genome dynamics and evolution of the mitochondrialand nuclear genomes in Saccharomyces sensu lato species. Ph.D. Thesis.University of Copenhagen, Denmark.

17b.Piskur, J. Unpublished data.18. Piskur, J. 1997. The transmission disadvantage of yeast mitochondrial inter-

genic mutants is eliminated in the mgt1 (cce1) background. J. Bacteriol.179:5614–5617.

19. Piskur, J., S. Smole, C. Groth, R. F. Petersen, and M. B. Pedersen. 1998.Structure and genetic stability of the mitochondrial genomes vary amongyeasts of the genus Saccharomyces. Int. J. Syst. Bacteriol. 48:1015–1024.

20. Ryu, S. L., Y. Murooka, and Y. Kaneko. 1996. Genomic reorganizationbetween two sibling yeast species, Saccharomyces bayanus and Saccharomy-ces cerevisiae. Yeast 12:757–764.

21. Spencer, J. F. T., and D. M. Spencer. 1977. Apparent bisexual behavior ofyeast strains obtained from hybridization of industrial yeasts of the genusSaccharomyces with auxotrophic diploids. Antonie van Leeuwenhoek 43:245–254.

22. Vaughan-Martini, A., A. Martini, and G. Cardinali. 1993. Electrotypickaryotyping as a taxonomic tool in the genus Saccharomyces. Antonie vanLeeuwenhoek 63:145–156.

23. Wolfe, K. H., and D. C. Shields. 1997. Molecular evidence for an ancientduplication of the entire yeast genome. Nature 387:708–713.

6496 MARINONI ET AL. J. BACTERIOL.

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from