Proc. Natl Acad. Sci. USAVol. 80, pp. 3035-3039, May 1983Genetics

Regulation of yeast mating-type interconversion: Feedback controlof HO gene expression by the mating-type locus

(genetic rearrangement/eukaryotic gene control/regulatory genes/cassettes)

ROBERT JENSEN*tt, GEORGE F. SPRAGUE, JR.*, AND IRA HERSKOWWIZ*tt*Institute of Molecular Biology and Department of Biology, University of Oregon, Eugene, Oregon 97403; and tDepartment of Biochemistry and Biophysics,University of California, San Francisco, California 94143

Communicated by Franklin W. Stahl, February 24, 1983

ABSTRACT The ultimate product of yeast mating-type in-terconversion is a stable a/a diploid cell. A haploid cell carryingthe HO gene gives rise to a diploid cell in a two-step process: first,the cell switches mating type as a result of genetic rearrangement(cassette substitution) catalyzed by HO; then, cells of opposite typemate to form a/a diploids. Mating-type interconversion does notoccur in a/a diploids despite the presence of the HO gene. Wehave identified a plasmid carrying the HO gene by screening ayeast clone bank (constructed in vector YEpl3) for plasmids thatallow mating-type switching by ho cells. The yeast segment re-sponsible for mating-type interconversion integrates by homologyat the ho locus, thus confirming that it carries HO. Using the HOgene as a probe, we find that strains with an active mating-typeinterconversion system produce HO RNA, whereas a/ca HO/HOcells do not and that this inhibition requires products of both theMATal and MATa2 genes. Thus, mating-type interconversion doesnot occur in a/a HO/HO cells because the HO gene product isnot synthesized. These results demonstrate the following: (i) Themating-type locus, proposed on genetic grounds to be a regulatorylocus, controls expression of an unlinked gene (HO) at the level ofRNA production. (ii) TheHO gene is under negative feedback con-trol: its expression is inhibited after successful completion of dip-loidization (formation of a/a diploids).

The HO locus of the yeast Saccharomyces cerevisiae governsthe frequency of mating-type interconversion (ref. 1; reviewedin ref. 2).§ Strains carrying HO switch between MATa and MATaas often as every cell division, whereas strains carrying the re-cessive ho allele switch at much lower frequency. HO promotesa site-specific, unidirectional transposition of a cassette of ge-netic information from the HML and HMR loci (where the cas-settes are silent) to the mating-type locus, where the a or a cas-sette is expressed and determines yeast cell type. As firstdescribed by Winge and Roberts (1), the HO gene promotesformation of diploid cells from haploids, a process termed "dip-loidization": haploid MATa or MATa cells carrying HO give riseto sibling cells of opposite mating type as a result of mating-type interconversion, and these cells then mate to form a/acells. Despite being homozygous for HO, the resultant MATa/MATa cells remain MATa/MATa diploids. The final productof mating-type interconversion is thus a stable a/a diploid de-rived from a haploid a or a cell.

Mating-type interconversion is controlled by the mating-typelocus: switching occurs in MATa/MATa and MATa/MATadiploids but not in MATa/MATa diploids (see ref. 3). Thus,stability of MATa/MATa HO/HO cells results not from diploi-dy per se but rather because of heterozygosity at the mating-type locus. More specifically, inhibition of mating-type inter-conversion in a/a cells requires the al function of MATa and

the a2 function of MATa: matal/MATa and MATa/mata2diploids carrying HO exhibit mating-type interconversion andswitch cassettes at MAT until the genome contains both MATaand MATa (3, 4). al and a2 of course are also necessary for themore familiar properties of a/a cells-sporulation proficiencyand inability to mate.One explanation for the inability of a/a HO/HO cells to ex-

hibit mating-type interconversion is that HO or some otherfunction necessary for mating-type interconversion is not ex-pressed in MATa/MATa cells. (A different type of hypothesisis discussed in ref. 3.) Genetic evidence indicates that the mat-ing-type locus controls expression of unlinked genes (Fig. 1)(5-7). Recent studies (8) confirm this hypothesis for at least onegene (STE3) that is unlinked to MAT and that is required formating: synthesis of STE3 RNA occurs only in a cells (and notin a or a/a cells) and requires the al product of MATa. Todetermine whether HO expression is also controlled by themating-type locus, we have cloned a genomic segment of yeastDNA that carries the HO gene and used it as a probe for expres-sion of HO in cells that differ in their ability to switch mating-type cassettes. We find that HO RNA is not produced in MATa/MATa HO/HO cells. Thus, HO represents a class of genesthat is expressed in haploid cells (or in diploids homozygous forMAT) but not in a/a diploid cells.

MATERIALS AND METHODSStrains and Relevant Genetic Markers. G150-15d (ho MATa

stel3-1 leu2-3 leu2-112; ref. 8), X10-lb (MATa/MATa diploidhomozygous for HO HMLa HMRa; ref. 9), 3B54 (a UV-inducedstel4 mutant derived from X10-lb; ref. 10), AB320 (MATa/MATa HO/HO; ref. 11), SC3 (MATa ho ura3-52; R. W. Davis'laboratory, obtained from T. Etcheverry), J12-1A (MATa/MATadiploid homozygous for HO ura3-1, obtained from J. Kurjan),341 (MATa/MATa diploid homozygous for HO HMLa HMRaura3), HR125-6d (ho MATa leu2-3 leu2-112 ura3-52 stel3-1),HR112-1b (HO MATa HMLa HMRa), HR100-1a (HO MATaHMLa HMRa), and 523 (a MATa/MATa mitotic recombinantderived from a MATa/MATa diploid homozygous forHO HMLaHMRa) were constructed for this work by standard methods.HR100-la and HR112-lb are very closely related to X10-lb:strains carrying the nonstandard alleles at HML or HMR (HMLa,

Abbreviations: bp, base pair(s); kb, kilobase(s); kbp, kilobase pair(s).t Present address: Dept. of Biochemistry and Biophysics, Univ. of Cal-ifornia, San Francisco, CA 94143.

§ Genetic nomenclature. MAT, HML, and HMR are genetic loci thatharbor either an a or an a cassette. Standard yeast strains carry an acassette at HML (HMLa) and an a cassette at HMR (HMRa). MATal,MATa2, and MATal are genes that code for products al, a2, and al,respectively. Mutations in MAT are denoted as mat; for example, a mu-tation in MATal is matal. Ho indicates the HO gene product or itsactivity.

3035

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 2,

202

0

Proc. Natl. Acad. Sci. USA 80 (1983)

MAT UNUNKED GENES

FIG. 1. Control of cell type by the mating-type locus (the al-a2 hy-pothesis). Panels show the structure and expression of the mating-typelocus (MAT) alleles for a, a, and a/a cells according to the al-a2 hy-pothesis (6). Unlinkedgeneswhose expression is controlledbyMAT aredrawn to the right. Wavy line indicates gene expression; line with anarrowhead indicates stimulation of gene expression; line with a ter-minal bar indicates inhibition ofgene expression. asg, a-specific genes;asg, a-specific genes; dsg, diploid-specific genes (genes required forsporulation and other functions that are expressed in a/a cells). Circledsymbols indicate the regulatory products that carry out stimulation orinhibition as described in the text.

HMRa) were obtained after UV irradiation of X10-lb (10); thesestrains were then backcrossed to X10-lb three times to yieldHR100-la and HR112-lb. XP11 (MATa/MATa ho/ho) was

formed by mating between XP8-18b (MATa ho) and a MATaho strain derived from XP8-18b by mating-type interconver-sion; XP8-18b is an Ade' revertant of XT1172-S245c (5); thesestrains were constructed by Peter Kushner. G67A3 (MATa/mata2-1 ho/ho) and G57A1 (MATa/matal-2 ho/ho) were

formed by matings between MATa strain XR28-29c and theoriginal mata2-1 and matal-2 derivatives of XT1172-S245c.G40 (matal/MATa ho/ho) was formed by mating betweenXT1172-S245c and strain 17-15 (12).

Media and Genetic Methods. Media and genetic methods(scoring HO, a-factor response of single cells) are described inref. 9. For mapping the site of integration of YIp5-B2, Ura'transformants of ho ura3 strain SC3 were mated with HO ura3strains and analyzed as described in the text. The stable Ura'transformants of SC3 were phenotypically Ho-. Apparently a

functional HO gene is not reconstituted by recombination be-tween ho and the B2 segment. Parental ditype tetrads from thesediploids were 2 Ho' Ura-:2 Ho- Ura'.

Identification of YEpHO. A clone bank in plasmid vectorYEpL3 (13) was constructed by insertion of yeast DNA frag-ments (from HO strain AB320) obtained after partial digestionwith endonuclease Sau3A as described in ref. 11 and kindlysupplied by K. Nasmyth in Escherichia coli. Plasmid DNA was

isolated and used to transform G150-15d (ho MATa stel3 leu2)to Leu+ as in ref. 14. Transformants were collected, replated,

and tested for the ability to mate as a by replica-plating withtester strain XR197b-la (matal HMLa HMRa sir-I leu2, whichmates efficiently with a cells). Plasmid DNA was isolated fromyeast as in ref. 11. Transformation of E. coli strains and sub-cloning were performed by standard methods.

Hybridization Analysis. Yeast RNA was isolated as follows(R. Elder, personal communication): 100-ml cultures of expo-nentially growing cells were harvested by centrifugation andwashed with sterile H20. Approximately 2 X 109 cells were sus-pended in lysis buffer (0.5 M NaCI/0.2 M Tris'HCl, pH 7.5/0.01 M EDTA/1% NaDodSO4) and added to 4 g of acid-washedglass beads (0.25-0.30 mm). Two milliliters of phenol/CHC13,1:1 (vol/vol), was added, and the mixture was vortexed for 2.5min. After addition of 3 ml of lysis buffer and 3 ml of phenol/CHC13, the extract was centrifuged to separate phases, and theaqueous phase was reextracted with phenol/CHC13. RNA wasprecipitated from the aqueous phase with ethanol. Poly(A)+ RNAwas isolated as in ref. 15 and fractionated on 1.5% agarose/6%formaldehyde as in ref. 16, except that Hepes buffer (pH 7.8)was substituted for borate buffer. RNA was transferred to ni-trocellulose and hybridized with probe as in ref. 17, except thatdextran sulfate was absent from hybridization solutions. Probewas prepared by nick-translation of plasmid DNA by using aNew England Nuclear nick-translation kit. Washed filters wereautoradiographed for 48 hr with Kodak XAR-2 film and a DuPont1 Lightning Plus intensifying screen.

Cloning a Segment Containing ho. A collection of EcoRI DNAfragments (from ho strain S288C) inserted into Agt (18) was ob-tained from R. W. Davis via F. W. Stahl and was screened withHO probe (obtained by nick-translation of plasmid YIp5-BH2)as in ref. 19. A 2.5-kilobase pair (kbp) HindIII fragment cor-responding to fragment H2 of YEpHO was inserted into plas-mid YIp5 and assayed for Ho activity as described in the text.

RESULTSIdentification and Properties of a Cloned DNA Segment

That Allows Mating-Type Interconversion in ho Cells. We haveidentified a plasmid that carries HO by screening for plasmidswith the ability to promote mating-type interconversion in hostrains. The ho recipient used in our transformations was MATaand carried a mutation in an a-specific STE gene (STE13; ref.7), a gene required for mating by a cells but not by a cells (seeTable 1). An ho MATa stel3 cell and the colony derived fromit are unable to give a mating response with either a or a testercells. In contrast, MATa stel3 cells that carry theHO gene switchto MATa; hence a cell that is initially MATa stel3 forms a col-ony that contains a mixture of two types of cells-those thatmate as a (genotypically MATa ste13) and others like the orig-inal cell that are unable to mate (genotypically MATa stel3) (ref.10; see also ref. 20). Therefore, we have screened for plasmids

Table 1. Phenotypes of ste13 strains carrying ho orHO allelesIndividual cell Colony,

Genotype of cell Mating a-Factor response matingMATa steM ho No No NoMATa stel3 HO* No No Yes (as a)t

MATa steM ho Yes (as a) Yes Yes (as a)MATa steM HO* Yes (as a) Yes Yes (as a)t

Mating and a-factor response by individual cells and by colonies grownfrom these cells were assayed as described in the text.*HO provided either at its standard chromosomal position or on a plas-mid (such as YEpHO).tThe[e colonies contain two types of cells: MATa steM andMATa ste3.

as cell &isg,FK asg

4 - _0

c2 acHO

dsg

acell osg

asgal -I

HO

dsg

4a/ cell Ad xsg4V2 cvi

Yr;--K atsgal

gHO

1I

3036 Genetics: Jensen et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 2,

202

0

Proc. Natl. Acad. Sci. USA 80 (1983) 3037

that allow ho MATa ste3 strains to switch to MATa.An ho MATa steW leu2 strain (G150-15d) was transformed

with a plasmid pool containing random genomic yeast DNAfragments inserted into the LEU2' vector YEp13. Of 20,000Leu+ transformant colonies screened for mating ability, 1 con-tained a plasmid (YEpHO) that allowed the recipient cells toform colonies that mated as a. Direct microscopic observationof individual cells confirmed that cells carrying YEpHO switchbetween two cell types: cells that were initially sensitive to afactor (phenotypically a, genotypically MATa steW) gave rise tocells that were resistant to a factor (phenotypically not a, geno-typically MATa stel). These latter cells then gave rise to cellsthat responded to a factor. Further evidence that YEpHO al-lowed mating-type interconversion of ho strains came from thefollowing observations: (i) MATa and MATa ho STE' cells car-rying YEpHO efficiently gave rise to nonmating, sporulation-proficient diploid cells-that is, to MATa/MATa cells. (ii)YEpHO allowed cells with a mutation of the mating-type locus(either matal or mata2) to switch readily to MATa and MATa,respectively, and thus to form MATa/MATa cells. The YEpHOplasmid, like a chromosomal HO, thus can heal mutations ofthe mating-type locus (3, 20). Demonstration that YEpHO car-ries HO is given in the next section.

To localize the region responsible for HO activity in the orig-inal 6-kbp insert in YEpHO, we subcloned restriction endo-nuclease fragments into the plasmid vector YIp5. When pres-ent in either orientation in YIp5, a 2.5-kbp HindIII fragment(denoted "H2"; Fig. 2) provided Ho activity when introducedinto ho MATa steW ura3 strain HR125-6d. Although YIp5 isincapable of autonomous replication (21), YIp5-H2 yielded un-stable transformants at a high efficiency. Thus, H2 apparentlycontains a sequence (an ARS sequence; ref. 22) that allows au-tonomous replication of YIp5. A smaller fragment (the 1. 1-kbpBamHI fragment, B2) provided Ho activity when inserted inone orientation in YIp5 but not in the other orientation. The870 base pair (bp) BamHI-HindIII fragment (BH2) represent-ing the overlap between the H2 and B2 fragments lacked Hoactivity in YIp5. (The only possible orientation of BH2 in YIp5is the same as the inactive orientation of B2 in the plasmid.)Taken together, these results indicate that most of the infor-mation for Ho activity is carried on the 870-bp BH2 fragment.YEpHO Carries the HO Gene. To determine whether the

insert in YEpHO carries the HO gene itself (and not another

AH2 2.5 kbp

H

BHI

B FRAGMENT

H2(1,). B2(I)

B H

BH2 0.87 kbp BH3

B2 1.1 kbp

Ho AC77VITY

B2(U), BH2 (II)

FIG. 2. Restriction map of cloned DNA fragment containing HO(A) and localization of Ho activity in subfragments (B). (A) Positions ofBamHI (B) andHindll (H) restriction endonuclease sites are drawn toapproximate physical scale. (B) Restriction endonuclease fragments H2,B2, and BH2 (as shown in A) were inserted into plasmid YIp5 and as-sayed for Ho activity as described in the text. Iand IIare orientationsof inserted fragments with respect to the unique HindIsand BamHIsites of YIp5. H2(I) is an insert of H2 and YIp5 such thatthe YIp5Hindssite is to the left and the BamHI site to the right of the H2 fragmentas drawn in A. All fragments shown except BH3 exhibited high-effi-ciency transformation, suggesting that they carry an ARS sequence.

gene normally controlled by HO or otherwise limiting for mat-ing-type interconversion), we allowed a plasmid containing theputative HO DNA to integrate into the yeast genome by re-combination with homologous sequences and determined itssite of insertion. If the cloned DNA segment contains HO, theplasmid should integrate at the HO locus (or its allele ho). Forthis analysis, we have used the URA3V plasmid YIp5 carryingthe B2 fragment. When transformed into a ura3 mutant, stableUra+ transformants arise by integration of the plasmid into thegenome (21). (Integration at the defective ura3 locus is greatlydecreased by using the ura3-52 mutation in the recipient; ref.23). Stable Ura' transformants obtained in recipient strain SC3(ho MATa ura3-52) were crossed to HO MATa ura3 spores (fromstrains J12-1A and 341), and meiotic products were analyzed.In crosses with three independently isolated Ura+ integrants,the Ura+ phenotype contributed by YIp5-B2 was tightly linkedto the ho locus (within 1 centimorgan): no recombinants wereobtained in a total of 87 tetrads (see Materials and Methods).In other crosses, we observed that the Ura+ determinant is linkedto CDC9 (data not shown) at the same frequency as is HO (G.Kawasaki, cited in ref. 24). These results show clearly that URJA3is now located at the ho locus and thus that the B2 segmentcontains nucleotide sequences present at this locus.

Additional evidence that YEpHO contains the HO gene comesfrom analysis of a genomic DNA segment containing the ho al-lele. A fragment from an ho strain was identified by homologywith the BH2 fragment of YEpHO (as described in Materialsand Methods), and a 2.5-kbp HindIII subfragment (H2', whichcorresponds to the H2 fragment of YEpHO) was inserted intoYIp5. YIp5-H2' does not promote mating-type interconversionwhen introduced into an ho MATa stel3 strain. The inabilityof this segment, derived from an ho strain, to promote mating-type interconversion provides an independent argument thatthe homologous segment present in YEpHO and derived froman HO strain carries the HO gene.

Control of HO RNA Synthesis by the Mating-Type Locus.To determine whether the mating-type locus controls expres-sion of the HO gene, we used the cloned HO gene as a probefor RNA isolated from HO strains that differ in their ability toswitch mating types. Because haploid HO cells diploidize toform MATa/MATa cells, a culture of haploid MATa or MATaHO cells cannot be used as a source of RNA for cells exhibitingmating-type interconversion. Therefore, we have used HO strainsin which the interconversion system remains active because thesestrains do not form a/a diploids. One strain (3B54) contains ana-specific STE mutation (stel4) that prevents mating betweendaughter cells (10). A second strain (HR112-lb) carries only acassettes at MAT, HML, and HMR and thus cannot form a/adiploids. Similarly, a third strain (HR100-la) contains only acassettes at MAT, HML, and HMR. The mating-type intercon-version system is active in such strains (25, 26). These strainswere compared with HO strains in which mating-type inter-conversion does not occur, MATa/MATa HO/HO diploids X10-lb (the parent of the three prior strains) and AB320 (from whichthe clone bank was derived). We note that X10-lb, HR112-lb,and HR100-la are isogenic (see Materials and Methods); AB320and 523 (see below) are from different backgrounds.

Poly(A)+ RNA was isolated from these strains and hybridizedwith a probe (YIp5-BH2) that contains URA3 and part of HO.As shown in Fig. 3, we observe a species of RNA [=1.5-1.7kilobases (kb)] complementary to the cloned DNA probe thatis present in the haploid MATa and MATa HO strains, in whichmating-type interconversion is active (Fig. 3, lanes A-C); thisspecies is completely absent in HO/HO MATa/MATa dip-loids, in which mating-type interconversion does not occur (Fig.3, lanes E and F). This RNA species is also present in diploid,HO/HO MATa/MATa cells that are homozygous for HMLa

Genetics: Jensen et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 2,

202

0

Proc. Natl. Acad. Sci. USA 80 (1983)

A B C D E F A B C D E F

-m w__ ___p

FIG. 3. Synthesis ofHO RNAbyHO strains differing in activity ofthe mating-type interconversion system. Poly(A)+ RNA was isolatedfrom strains 3B54 (lane A; mixed population of MATa stel4 HO andMATa stel HO), HR112-lb (lane B; MATaHMLaHMRa HO), HR100-la (lane C; MATa HMLa HMRa HO), 523 (lane D; MATa/MATaHMLa/HMLaHMRa/HMRa HO/HO), X10-lb (lane E, MATa/MATaHMLa/HMLa HMRa/HMRa HO/HO), and AB320 (lane F; MATa/MATaHMLa/HMLaHMRa/HMRaHO/HO). RNA was fractionatedand transferred to nitrocellulose paper. Immobilized RNA was hybrid-ized with 32P-labeled, nick-translated YIp5-BH2 plasmid DNA, whichcontains URA3 and part ofHO, and autoradiographed.

and HMRa (Fig. 3, lane D). Al strains produce similar amountsof URA3 RNA, which is 0.9-1.0 kb (27). These results show aperfect correlation between active mating-type interconversionand production of a RNA that is homologous to a DNA segmentcarrying HO. Thus, mating-type interconversion does not occurin a/a cells because HO RNA is not produced. Whether tran-scription of HO is blocked in a/a cells or the HO transcript isunstable cannot be determined from our analysis.

Because the HO and URA3 segments of the probe are of ap-proximately equal size, the extent of hybridization to the HOand URA3 bands indicates the relative amounts of these RNAspecies. These bands were excised from the nitrocellulose filterand radioactivity was measured in a liquid scintillation counter,which showed the HO to URA3 ratio to be 41:2. Thus, the levelof stable HO transcript appears to be rather low, only half thatof URA3 (which is present in -5-10 molecules per cell; ref.27). [We note that 2.5-5 HO transcripts per cell is an average:only certain cells within a clone of HO cells are competent toswitch mating types (9, 28), and it is perhaps only these cellsthat express HO.] By similar quantitative analysis, we find noHO transcript in a/a cells (that is, no hybridization above back-ground). Thus, production of the HO transcript in a/a cells is<1% of the level in expressing cells.As noted above, physiological studies show that both a2 and

al products are necessary to inhibit mating-type interconver-sion. Thus, we anticipated that these products would also berequired for inhibiting synthesis of HO RNA. Because MATa/mata2 and matal/MATa strains carrying HO switch mating-type cassettes efficiently, it is not possible to grow pure cultures

with these genotypes to assay HO RNA in such strains. How-ever, we have found that ho strains produce a transcript whose

FIG. 4. Regulation of synthesis of RNA from the ho locus by themating-type locus. Poly(A)+ RNA was isolated from different strainsand hybridized to YIp5-BH2 probe. Lanes: A, ho/ho MATa/MATa(strain XP11); B, ho MATa (XP8-18b); C, mixed population ofHOMATasteW andHOMATa steW (3B54); D, ho/ho MATa/mata2-1 (G67A3);E, ho/ho matal/MATa (G40); and F, ho/ho MATa/matal-2 (G57A1).Strains XP11 and XP8-18b (lanes A and B) are isogenic, as are strainsG67A3 and G57A1 (lanes D and F) (see Materials and Methods).

size is identical to that produced by HO strains (Fig. 4, lanesB and C) and that this transcript is controlled in the same man-ner as in HO strains, expressed in haploids and not in MATa/MATa diploids (Fig. 4, lanes A and B). This RNA is absent fromMATa/matal strains (Fig. 4, lane F) and present in MATa/mata2 and matal/MATa strains (Fig. 4, lanes D and E). Thus,a2 and al are required in diploid cells to inhibit production ofRNA from the ho locus. The functional significance of the tran-script produced in ho strains is unknown.

DISCUSSION

HO and the mating-type locus exert a mutual control over eachother: the HO gene product is responsible for switching allelesof the mating-type locus, and the mating-type locus regulatesactivity of the HO gene. The latter observation has two mainimplications: first, the mating-type locus controls expression ofan unlinked gene at the level of RNA production; second, mat-ing-type interconversion does not occur in a/a cells becausethe HO gene product is not synthesized.The essential components of the al-a2 model for control of

yeast cell type are that the mating-type locus controls expres-sion of genes unlinked to the mating-type locus and that it doesso by specifying three regulatory activities (6). al (coded by theMATal gene) is a positive regulator of at least some a-specificgenes; a2 (coded by the MATa2 gene) is a negative regulatorof at least some a-specific genes (see Fig. 1). The action of clrand a2 in a cells, and their absence in a cells, accounts for theproperties of a and a cells. The properties of a/a cells are dic-tated by two regulatory activities, a2 acting by itself and a2acting with al (coded by the MATa gene) to form "al-a2." a2alone acts just as it does in haploids, to inhibit expression of a-specific genes. al-a2 inhibits expression of the MATal gene

..A..AMNMMIL-Ammw. --

V.qwmw qm..

3038 Genetics: Jensen et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 2,

202

0

Proc. Natd Acad. Sci. USA 80 (1983) 3039

(29, 30); hence, a-specific genes are not expressed. The levelof control exerted by these regulators is known to be at the levelof RNA production for two cases: al is required for synthesisof RNA from the a-specific STE3 gene (8); and al-a2 nega-tively regulates synthesis of RNA from the MATal gene (29,30). We have shown here that al-a2 negatively regulates theHO gene as well. This conclusion comes from the following ob-servations: (i) MATa/MATa and MATa/MATa (and corre-spondinghaploids) but not MATa/MATa cells produce HO RNA.(ii) matal/MATa and MATa/mata2 strains produce ho RNA.Thus, it is clear that both al and a2 are required to inhibitexpression of the ho (HO) locus. Haploid strains that are phe-notypically a/a because they express a and a cassettes at HMLand HMR (due to a mutation allowing expression of these cas-settes), as expected, do not exhibit mating-type interconversion(31) and do not express HO (unpublished data).

al and a2 inhibit expression of several genes in addition toHO and MATal. Production of RNA from the STE5 gene (re-quired for mating by both a and a cells) is inhibited in a/a dip-loids (V. L. MacKay, J. Thorner, and K. Nasmyth, personalcommunication). Thus, HO and STE5 are genes that can betermed "haploid-specific," expressed in haploid strains (thatare phenotypically a or a) but not in diploid strains (that arephenotypically a/a). Production of RNA encoded by the re-peated element Tyl is likewise inhibited in a/a cells (32). As-sociation of a Tyl element with several different loci (CYC7,DUR, ADR2) places these loci under MAT control: expressedat a higher level in haploids than in a/a diploids (33, 34). Fi-nally, we have proposed (35) that sporulation is triggered in a/a cells because al-a2 inhibits synthesis of a negative regulatorof sporulation.How does al-a2 exert its negative regulatory activity at so

many different, widely dispersed loci? Obviously, these loci mayshare a common recognition site for al-a2. Because geneticrearrangements that place Tyl adjacent to various genes causethese genes to be under al-a2 control, Errede et al. (33) havesuggested that Tyl may be naturally associated with certain yeastgenes and be responsible for their control by al-a2. We haveno evidence that HO is associated with such a Tyl element: thecloned HO segment does not hybridize to Tyl (unpublisheddata).

Diploidization is a two-step process, mating-type intercon-version followed by mating between haploid cells of oppositecell type. If mating-type interconversion were to continue inMATa/MATa HO/HO cells (to produce MATa/MATa andMATa/MATa cells), mating between siblings would result incells of ever-increasing ploidy-for example, production of a/a/a/a tetraploids. However, MATa/MATa cells are stable inthe presence of HO because, as shown here, the HO geneproduct is not synthesized. (We note that mating-type inter-conversion is turned off immediately after formation of an a/a HO/HO cell; hence, some component of the mating-typeinterconversion machinery, perhaps the HO gene product it-self, may be unstable.) al-a2 registers the successful formationof an a/a cell and prevents further mating-type interconver-sion. The a/a diploid state is stable until meiosis and sporu-lation yield haploid spores, which once again express the HOgene and resume mating-type interconversion and diploidiza-tion. Thus, diploidization in yeast demonstrates one mecha-nism-intracellular feedback control-by which the successfulexecution of a complex event is monitored. By contrast, suc-cessful formation of a A lysogen (another complex event in-volving genetic rearrangement) is optimized in a different man-ner, in this case, by prognostication (36). It is clear that geneticrearrangement is under the types of control that allow it to playa role in a variety of biologically important processes (37, 38).

We give special thanks to Drena Larson and Diane Morton for initialguidance in recombinant DNA methods, to Kim Nasmyth for gener-ously supplying us with his clone bank and YEpL3, to those cited fortheir personal communications, and to James Darnell, Ken Krauter,and Michael Thomashow for timely advice on RNA hybridization anal-ysis. We also thank Kevin Struhl for YIp5, Stan Fields for commentson the manuscript, Kerrie Rine for preparation of the figures, and Vir-ginia Staubach for preparation of the manuscript. This work was sup-ported by a Research Career Development Award (AI-00163) and Re-search Grant (AI-13462) from the National Institutes of Health to I.H.and by a Public Health Service National Research Service Award inGenetics to the Department of Biology.1. Winge, 0. & Roberts, C. (1949) C. R. Trav. Lab. Carlsberg Ser.

Physiol 24, 341-346.2. Herskowitz, I. & Oshima, Y. (1981) in Molecular Biology of the

Yeast Saccharomyces: Life Cycle and Inheritance, eds. Strathern,J. N., Jones, E. W. & Broach, J. R. (Cold Spring Harbor Labo-ratory, Cold Spring Harbor, NY), pp. 181-209.

3. Strathern, J. N., Blair, L. C. & Herskowitz, I. (1979) Proc. NatiAcad. Sci. USA 76, 3425-3429.

4. Klar, A. J. S., Fogel, S. & Radin, D. N. (1979) Genetics 92, 759-776.

5. MacKay, V. L. & Manney, T. R. (1974) Genetics 76, 273-288.6. Strathern, J., Hicks, J. & Herskowitz, I. (1981) J. Mol. Biol 147,

357-372.7. Sprague, G., Jr., Rine, J. & Herskowitz, I. (1981)J. Mol. Biol 153,

323-335.8. Sprague, G. F., Jr., Jensen, R. & Herskowitz, I. (1983) Cell 32,

409-415.9. Hicks, J. B. & Herskowitz, I. (1976) Genetics 83, 245-258.

10. Blair, L. C. (1979) Dissertation (Univ. of Oregon, Eugene, OR).11. Nasmyth, K. A. & Reed, S. I. (1980) Proc. Nati Acad. Sci. USA

77, 2119-2123.12. Kassir, Y. & Simchen, G. (1976) Genetics 82, 187-202.13. Broach, J. R., Strathern, J. N. & Hicks, J. B. (1979) Gene 8, 121-

133.14. Beggs, J. D. (1978) Nature (London) 275, 104-108.15. Lindberg, V. & Persson, T. (1972) Eur. J. Biochem. 31, 246-254.16. Derman, E., Krauter, K., Walling, L., Weinberger, C., Ray, M.

& Darnell, J. E., Jr. (1981) Cell 23, 731-739.17. Thomas, P. S. (1980) Proc. Natl. Acad. Sci. USA 77, 5201-5205.18. St. John, T. P. & Davis, R. W. (1979) Cell 16, 443-452.19. Benton, W. D. & Davis, R. W. (1977) Science 196, 180-182.20. Hicks, J. B. & Herskowitz, I. (1977) Genetics 85, 373-393.21. Struhl, K., Stinchcomb, D. T., Scherer, S. & Davis, R. W. (1979)

Proc. Natl Acad. Sci. USA 76, 1035-1039.22. Stinchcomb, D. T., Struhl, K. & Davis, R. W. (1979) Nature

(London) 282, 39-43.23. Scherer, S. & Davis, R. W. (1979) Proc. Natl Acad. Sci. USA 76,

4951-4955.24. Mortimer, R. K. & Schild, D. (1980) Microbiol Rev. 44, 519-571.25. Rine, J., Jensen, R., Hagen, D., Blair, L. & Herskowitz, I. (1981)

Cold Spring Harbor Symp. Quant. Biol 45, 951-960.26. Klar, A. J. S., Hicks, J. B. & Strathern, J. N. (1981) Cold Spring

Harbor Symp. Quant. Biol 45, 983-990.27. Rose, M. (1982) Dissertation (Massachusetts Institute of Tech-

nology, Cambridge, MA).28. Strathern, J. N. & Herskowitz, I. (1979) Cell 17, 371-381.29. Klar, A. J. S., Strathern, J. N., Broach, J. R. & Hicks, J. B. (1981)

Nature (London) 289, 239-244.30. Nasmyth, K. A., Tatchell, K., Hall, B. D., Astell, C. & Smith, M.

(1981) Nature (London) 289, 244-250.31. Klar, A. J. S., Strathern, J. N. & Hicks, J. B. (1981) Cell 25, 517-

524.32. Elder, R. T., St. John, T. P., Stinchcomb, D. T. & Davis, R. W.

(1981) Cold Spring Harbor Symp. Quant. Biol 45, 581-584.33. Errede, B., Cardillo, T. S., Sherman, F., Dubois, E., Des-

champs, J. & Wiame, J.-M. (1980) Cell 25, 427-436.34. Williamson, V. M., Young, E. T. & Ciriacy, M. (1981) Cell 23, 605-

614.35. Rine, J., Sprague, G., Jr., & Herskowitz, I. (1981) Mol Cel. Biol

1, 958-960.36. Herskowitz, I. & Hagen, D. (1980) Annu. Rev. Genetics 14, 399-

445.37. McClintock, B. (1968) Dev. Bido, Suppl. 1, 84-112.38. Campbell, A. (1981) Cold Spring Harbor Symp. Quant. BidL 45,

1-9.

Genetics: Jensen et aL

Dow

nloa

ded

by g

uest

on

Nov

embe

r 2,

202

0